home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Network Support Library
/
RoseWare - Network Support Library.iso
/
pressgen
/
laninf.exe
/
MANUAL.TRN
Wrap
Text File
|
1993-06-10
|
654KB
|
10,891 lines
1.0 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . G066881
1.1 A Comparison of LANs and PABXs
1.2 LAN Capabilities
1.3 Planning for a LAN
2.0 LAN TECHNOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . G066880
2.1 Physical LAN Attachment
2.1.1 Media Types
2.1.1.2 Fiber Optic Cable
2.1.2 Network Topologies
2.1.3 Transmission Techniques
2.2 LAN Medium Access Protocols
2.2.1 Basic CSMA/CD Concepts
2.2.2 Basic Token-Passing Ring Concepts
2.2.2.1 MAC Addressing
2.2.2.2 Early Token Release
2.2.2.3 Token Monitoring
2.2.2.4 Neighbor Notification
2.2.3 Basic Token-Passing Bus Concepts
2.3 Fiber Distributed Data Interface Concepts
2.3.1.2 FDDI Standards
2.4 LAN Station Management
2.4.1 Open Systems Management
2.4.1.2 Registration
2.4.2 Resource Management
2.4.3 Name Management
3.0 LAN ARCHITECTURES AND STANDARDS . . . . . . . . . . . . . . . . . G066879
3.1 LAN and Communications Standards
3.1.1 IEEE 802 and OSI
3.1.2 OSI and SNA Models
3.2 LAN Standards
3.2.1 Physical Layer and MAC Sublayer
3.2.1.1 IEEE 802.3, IS 8802-3
3.2.1.2 IEEE 802.4, ISO 8802-4
3.2.1.3 IEEE 802.5, ISO IS 8802-5
3.2.1.4 Fiber Distributed Data Interface
3.2.2 Logical Link Control Sublayer
3.2.2.2 IEEE 802.2, ISO 8802-2
3.2.2.3 LLC Protocol Data Unit
3.3 LAN Standards Summary
4.0 STRUCTURED WIRING . . . . . . . . . . . . . . . . . . . . . . . . G066878
4.1.1 Star-Wired Environments
4.1.2 Other Wiring Environments
4.2 IBM Cabling System
5.0 IBM LOCAL AREA NETWORK SOLUTIONS . . . . . . . . . . . . . . . . G066877
5.1 IBM Token-Ring Network
5.1.1 IBM Token-Ring Network Components
5.1.1.1 Cabling Components
5.1.1.2 IBM 8230 Controlled Access Unit
5.1.1.3 Repeaters and Converters
5.1.1.4 IBM 8218 Copper Repeater
5.1.1.5 IBM 8219 Optical Fiber Repeater
5.1.1.6 IBM 8220 Optical Fiber Converter
5.1.2 IBM Token-Ring Network Devices
5.1.3 IBM PC Network (Broadband) Components
5.1.4 IBM PC Network (Broadband) Interfaces
5.2 IBM PC Network Baseband
5.2.1 PC Network Baseband Components
5.2.2 PC Network Baseband Interfaces
5.3 Ethernet/IEEE 802.3
5.3.1 Workstations
5.3.2 Bridges
5.3.3 Gateways
5.4 Industrial LAN
5.4.1 The 8232 LAN Channel Station and Industrial LAN
5.4.2 Migration to MAP 3.0
5.4.3 MAP - SNA Gateway
5.4.4 Series/1 and Industrial LAN
5.5 TCP/IP
5.5.1 Structure
5.5.2 Addressing and Routing
5.5.3 TCP/IP Applications
5.6 IBM LAN Support Program
5.6.1 Base Programming Interfaces
5.6.1.1 Open IEEE 802.2 LLC Interface
5.6.1.2 NetBIOS Interface
5.6.1.3 LU 6.2 Interface
5.6.2 LAN Support Program Structure
5.6.3 OS/2 Extended Edition 1.1 LAN Support Structure
5.6.4 OS/2 EE 1.1 LAN Support versus IBM LAN Support Program V1.10
5.6.5 IBM LAN Support Program Version 1.2
5.6.6 OS/2 Extended Edition 1.2 LAN Support
5.7 3174-Peer Communications RPQ 8Q0718
5.7.1 General description
5.7.2 Customization and Management
5.7.3 Hardware/Software Requirements
5.7.4 Summary
5.8 IBM LAN Offerings Summary
6.0 LAN SEGMENTS INTERCONNECTION . . . . . . . . . . . . . . . . . . G066876
6.1 Bridges, Routers and Gateways
6.2 Bridged LANs
6.2.1 Bridge Configurations
6.3 Bridge Standards
6.3.1 Transparent Bridging
6.3.2 Source Routing
6.3.3 Source Routing/Transparent Bridging Inter-operability
6.4 Bridge LAN Management Interface
6.5 Remote Bridge
6.6 IBM LAN Bridges
6.6.1 IBM Token-Ring Bridges
6.6.2 IBM Token-Ring Network Bridge Program Version 2.2
6.6.3 IBM PC Network Bridge Program
6.7 IBM 8209 LAN Bridge
6.7.1 Token-Ring to Token-Ring
6.7.2 Token-Ring to Ethernet/IEEE 802.3
6.7.3 General Description
6.7.4 8209 Operation
6.7.5 Frame Conversion
6.7.6 The Utility Program
6.7.7 Configuration Switches
6.7.8 Token-Ring Management Support
6.7.9 Enhanced Ethernet Attachment Feature
6.7.10 Summary
6.8 IBM Bridge Products Coexistence and Migration
6.9 Routers
6.9.1 The IBM LAN to LAN Wide Area Network Program
7.0 LAN GATEWAYS . . . . . . . . . . . . . . . . . . . . . . . . . . G066875
7.1 PC/PS2 Gateways
7.1.1 SNA 3270
7.1.1.2 Personal Communications/3270
7.1.2 Asynchronous Communication PC Gateways
7.2 Dedicated SNA Gateways
7.2.1 SNA System/370 Host Connectivity
7.2.1.1 9370 Processor
7.2.1.2 37xx Communication Controllers
7.2.2 3174 Local Gateway
7.2.3 3174 Remote Gateways
7.2.4 Gateway INN/BNN Connectivity
7.3 Non-SNA Gateways
7.3.1 8232 LAN Channel Station
7.3.2 IBM 3172 Interconnect Controller
7.3.2.1 IBM 3172 Model 001
7.3.2.2 Interconnect Controller Program Version 2.0
7.3.2.3 Interconnect Controller Program Version 2.1
7.3.3 IBM 3172 Model 002
7.3.4 Interconnect Controller Program Version 2.x (Statement of
Direction)
7.3.5 IBM 3172 Interconnect Controller Summary
7.4 Host Gateways - Conclusion
7.4.1 Selection Criteria
7.4.2 Functional Comparison
7.4.3 Gateway Positioning
7.5 AS/400 LAN Attachment
7.6 3174 x3R Establishment Controllers
7.6.1 Single-link, multiple host support
8.0 LAN SERVERS . . . . . . . . . . . . . . . . . . . . . . . . . . . G066874
8.1 IBM PC LAN Program V1.2
8.1.2 IBM PC LAN Program V1.3
8.1.3 OS/2 LAN Server Version 1.2
8.1.4 System/36 PC LAN Support
8.1.5 AS/400 PC LAN Support
8.1.6 LAN PrintManager
9.0 LAN MANAGEMENT AND RECOVERY . . . . . . . . . . . . . . . . . . . G066873
9.1 IBM's Total Network Management Strategy
9.2 LAN Management and MAC Architecture
9.3 IBM LAN Management Products Overview
9.3.1 IBM PC 3270 Emulation LAN Management Program Version 1.0
9.3.2 IBM LAN Manager Entry Version 1
9.3.3 IBM LAN Manager V1.0
9.3.4 IBM LAN Manager V2.0
9.3.5 IBM LAN Network Manager V1
9.3.5.2 Support for the IBM 8230 Controlled Access Unit (CAU)
9.3.5.3 Security (Access Control)
9.3.6 IBM LAN Network Manager Entry
9.3.7 IBM LAN Network Manager V1.1
9.3.8 Migration and Coexistence
9.4 Management Capabilities of IBM SNA Gateways
9.4.1 3174
9.4.2 Communication Controllers
9.4.3 9370
9.4.4 AS/400
9.5 IBM LAN Station Manager V1.0
9.5.1 Name Management
9.5.2 Resource Management
9.5.3 Service Interface
9.5.4 Summary
9.6 NetView/PC
9.7 IBM Token-Ring Network Trace and Performance Facilities
9.8 IBM LAN Management Products Summary
10.0 LAN POSITIONING . . . . . . . . . . . . . . . . . . . . . . . . G066872
10.1 IEEE 802.3, ISO IS 8802-3
10.2 IEEE 802.4, ISO IS 8802-4
10.3 IEEE 802.5, ISO IS 8802-5
10.4 ANSI X3T9.5 ISO 9314 FDDI
10.5 LAN Interfaces
LIST OF ABBREVIATIONS
INDEX
END-OF-TOC
INTERNATIONAL TECHNICAL SUPPORT CENTER
LOCAL AREA NETWORK LIBRARY
LOCAL AREA NETWORKS
CONCEPTS AND PRODUCTS
Document Number GG24-3178-2
September 1990
International Technical Support Center
Raleigh, North Carolina
+--- TAKE NOTE --------------------------------------------------------------+
] ]
] Before using this information and the product it supports, be sure to read ]
] the general information under "Special Notices." ]
+----------------------------------------------------------------------------+
THIRD EDITION (SEPTEMBER 1990)
This edition applies to various IBM products available for current LAN
architectures.
Order publications through your IBM representative or the IBM branch office
serving your locality. Publications are not stocked at the address given
below.
A form for reader's comments appears at the back of this publication. If the
form has been removed, address your comments to:
IBM Corporation, International Technical Support Center
Dept. 985, Building 657
P. O. Box 12195
Research Triangle Park, NC 27709 USA
(c) Copyright International Business Machines Corporation 1990
Special Notices
This publication is intended to provide an understanding of LANs and IBM LAN
solutions for planning and support purposes. It contains descriptions of LAN
architectures, bridges, and LAN network management. The information in this
publication is not intended as the specification of the programming interfaces
that are provided by the IBM Token-Ring Network for use by customers in
writing programs that request or receive its services. See the PUBLICATIONS
section of the IBM PROGRAMMING ANNOUNCEMENT for the IBM Token-Ring Network.
References in this publication to IBM products, programs or services do not
imply that IBM intends to make these available in all countries in which IBM
operates. Any reference to an IBM product, program, or service is not
intended to state or imply that only IBM's product, program, or service may be
used. Any functionally equivalent program that does not infringe any of IBM's
intellectual property rights may be used instead of the IBM product, program
or service.
Information in this book was developed in conjunction with use of the
equipment specified, and is limited in application to those specific hardware
and software products and levels.
IBM may have patents or pending patent applications covering subject matter in
this document. The furnishing of this document does not give you any license
to these patents. You can send license inquiries, in writing, to the IBM
Director of Commercial Relations, IBM Corporation, Purchase, NY 10577.
The information contained in this document has not been submitted to any
formal IBM test and is distributed AS IS. The use of this information or the
implementation of any of these techniques is a customer responsibility and
depends on the customer's ability to evaluate and integrate them into the
customer's operational environment. While each item may have been reviewed by
IBM for accuracy in a specific situation, there is no guarantee that the same
or similar results will be obtained elsewhere. Customers attempting to adapt
these techniques to their own environments do so at their own risk.
Reference to PTF numbers that have not been released through the normal
distribution process does not imply general availability. The purpose of
including these reference numbers is to alert IBM customers to specific
information relative to the implementation of the PTF when it becomes
available to each customer according to the normal IBM PTF distribution
process.
The following products and features referenced in this publication are
trademarks of the International Business Machines Corporation:
Application System/400, AS/400
ES/3090, 3090, ES/9370, System/360, System/370
MVS/SP, MVS/XA
NCP
ACF/VTAM
NetView, NetView/PC
Operating System/2, OS/2
Operating System/400, OS/400
Personal Computer XT, PC/XT
Micro Channel
System/36, System/38
IBM
SAA
Presentation Manager
Personal Computer AT, AT
Personal System/2, PS/2
PROFS
ProPrinter
RT Personal Computer, RT PC, RT
The following terms are trademarks of other companies.
Ethernet is a trademark of Xerox Corporation
Unix is a trademark of AT & T.
Abstract
Document ID G066876
TITLE: GG24-3178
IN GREATER DETAIL IN "6.0 LAN SEGMENTS INTERCONNECTION."
in greater detail in "6.0 LAN Segments Interconnection."
Using a LAN segment as an elementary building block permits design of
interconnected multi-segment local area network with a wide range of possible
topologies. Again, each of these topologies has its advantages and
disadvantages. They are discussed in "6.2.1 Bridge Configurations."
Figure 7 under heading "2.1.2.1.4 Ring Topologies" introduces various
concepts related to multi-segment networks.
+------------------------------------------------------------------------------
]
]
] *---* *-----------* *-----------* ]
] ] ]---] ] ] ] ] *---*
] *---* ] ] *---* ] ] Bridge or ]---] ]
] --* ] ]---] ]---] ] Router ] *---*
] ] ] LAN ] *---* ] LAN ] *---* ] LAN
] W ] ] segment ] Bridge ] segment ]---] ]---] segment
] ] Gateway ] A ] 1 ] B ] *---* ] C
] A ]--/ *---* ] ] *---* ] ] ] *---*
] ] /--] ]---] ]---] ]---] ] ]---] ]
] N ] *---* ] ] *---* ] ] *---* ] *---*
] ] ] ] Bridge ] ]---] ] ]
] --* *-----------* 2 *-----------* *---*
]
] <--------------- Multi-segment Local Area Network -------------->
]
]
+------------------------------------------------------------------------------
Figure 7. Multi-Segment LAN Example
2.1.3 Transmission Techniques
2.1.3.1.1 Baseband Transmission: Baseband transmission involves putting a
signal directly on the transmission medium without modulating a carrier
signal. At any point in time, the information occupies the entire bandwidth
available on the medium. Sharing of the medium can be achieved by allocating
time slices to different applications. This technique is known as TIME
DIVISION MULTIPLEXING (TDM). It is a simple technique and together with the
fact that no modulating/demodulating devices are required, Baseband is a
relatively inexpensive transmission technology.
Baseband-transmitted signals must be regenerated periodically when
transmitting signals over long distances or through multiple connection
points. This is due to cable attenuation (losses in signal strength caused by
the physical and electrical characteristics of the medium) and losses
associated with connectors and/or attachments.
The IBM Token-Ring Network uses baseband transmission, operating at a speed of
4 or 16 Megabits per second (4 Mbps or 16 Mbps), and the IBM PC Network
Baseband uses baseband transmission at 2 Mbps.
2.1.3.1.2 Broadband Transmission: Broadband transmission involves use of
techniques to allow generation of signals over multiple channels. In most
cases, broadband transmission is based upon use of radio frequency modems in
the same way as radio and television signals are transmitted. Resource
sharing is accomplished by dividing the medium into logical channels using
FREQUENCY DIVISION MULTIPLEXING. Any individual channel could use time
division multiplexing to provide further sharing.
Since broadband transmission uses multi-channel "CATV-like" technology, it is
ideally suited to transmit voice and video. Hence, the broadband approach
offers the greatest potential to support voice, video, and data on a single
cabling system today. For this reason broadband transmission is quite popular
in manufacturing plants.
Transmission of video signals requires only a single channel, because it is
transmit only. Data communication involves both transmit and receive.
Therefore two channels are required. In order to minimize interference between
the transmit and receive channels, significantly separated frequency ranges
should be used. Within a transmit/receive channel pair, this characteristic is
indicated as the frequency offset, with the forward (higher) frequency range
separated from the return (lower) frequency range by a set of frequencies
referred to as the guard band. Within each range a sub-range of 6 Mhz defines
one channel of a channel pair, where the pair is composed of a return channel
below the offset, and a forward channel which corresponds to the lower channel
frequency plus the offset. See Figure 8 under heading "2.1.3.1.2 Broadband
Transmission".
+------------------------------------------------------------------------------
]
]
] ] ]1]1]2]2]3]4]4] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
]Frequency ]5]1]7]3]9]5]1]7] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
]MHz ].].].].].].].].] ] ] ] ] ] ] ] ]1]1]1]1]1]1]1]1]1]1]1]1]1]1]1]1]1]
] ]7]7]7]7]7]7]7]7]5]6]6]7]7]8]9]9]0]0]1]2]2]3]3]4]5]5]6]6]7]8]8]9]9]
] ]5]5]5]5]5]5]5]5]4]0]6]2]8]4]0]6]2]8]4]0]6]2]8]4]0]6]2]8]4]0]6]2]8]
]------------------------------------------------------------------------------
] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
]Standard ]T]T]T]T]T]T]T] ]2]3]4]4]5]6]F]F]F] ] ]A]B]C]D]E]F]G]H]I]7]8]9]1]1]
]CATV ]7]8]9]1]1]1]1] ] ] ] ]A] ] ]M]M]M] ] ] ] ] ] ] ] ] ] ] ] ] ] ]0]1]
]Channels ] ] ] ]0]1]2]3] ] ] ] ] ] ] ]1]2]3] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] ]------------------------------------------------------------------
]Frequency ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
]Bands ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] Sub-Split ]R]R]R]R]*]*]*]*]*]F]F]F]F] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] (4 Pairs)] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] Mid-Split ]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]*]*]*]*]*]*]*]*]*]*]F]F]F]F]F]F]
] (17 Pairs)] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] High-Split]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]R]*]*]*]
] (30 Pairs)] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
] -------------------------------------------------------------------
]
]R = Return Band
]F = Forward Band
]* = Guard Band
]
+------------------------------------------------------------------------------
Figure 8. Broadband Transmission Frequency Bands and Offsets
With respect to the number of channels and their associated guard bands,
industry standard channel pairs have been defined with specific frequency
offsets between the channels within a pair. Mid-split broadband transmission
refers to 17 channel pairs and a 116 to 168 MHz guard band (168 Mhz offset).
High-split refers to 30 channel pairs and a guard band of 186 to 222 MHz (122
Mhz offset).
Due to the high cost involved with the multi-channel capability, a modulated
transmission technique on a single channel pair medium has been defined and is
referred to as carrierband transmission.
The IBM PC Network (Broadband) uses broadband transmission at 2 Mbps,
supporting one of the following channel pair frequency options:
o Return Band frequency 50.75 Mhz, Forward Band 219 Mhz
o Return Band frequency 56.75 Mhz, Forward Band 249 Mhz
o Return Band frequency 62.75 Mhz, Forward Band 255 Mhz.
2.2 LAN Medium Access Protocols
2.2.1 Basic CSMA/CD Concepts
The access to the medium on a LAN controlled by a CARRIER SENSE MULTIPLE
ACCESS WITH COLLISION DETECTION (CSMA/CD) access protocol is based on
contention between competing stations. The access is said to be PROBABILISTIC
rather than deterministic. CSMA/CD is appropriate for use only with bus or
tree topologies because of the collision detection mechanism. In order to
detect signal collisions, all stations on the LAN must be able to sense the
presence of a carrier on the medium. This does not happen with a ring
topology in which a signal is passed serially from one station to the next.
Three main topology/transmission technique combinations are used for CSMA/CD
LANs: baseband bus (the most popular), baseband bus with star-wired device
controllers, and broadband tree. In each case, station attachment is passive.
The signal is neither regenerated nor amplified by the LAN stations. This
limits the maximum non-repeated station-to-station distance (typically 500
meters on a baseband bus). In addition, distance between attaching stations
must be a minimum of 2.5 meters to minimize the effect of signal reflections
in the taps (that is to avoid reflections adding up in phase).
Before transmitting data, a station first listens to the medium to find out
whether the medium is idle or whether another station is already transmitting
data. If the network is quiet, the station may start its own transmission.
The protocol provides multiple access, that is the access mechanism applies to
all LAN stations concurrently. Therefore it is possible that two stations
could start to transmit at about the same time. Due to propagation delays
(media delay, delays at taps and/or repeaters), a station may determine that
the medium is idle and start to transmit, while at the other end of the bus a
station has already started transmission. Two (or more) stations concurrently
broadcasting data on a common medium inevitably cause a collision. The
collision is detected as the station listens to the medium while it is
transmitting (COLLISION DETECTION). This process is called LISTEN WHILE TALK.
Both stations stop transmission and try again later.
The critical part of the CSMA/CD protocol is the recovery process after a
collision has occurred. It is this aspect of the CSMA/CD medium access
protocol that causes response time and throughput to be probabilistic.
In general, collision recovery proceeds as follows:
1. As soon as a station detects a collision, it stops transmitting data
immediately and sends a jamming signal. Since the first condition to make
any recovery possible is to stop all additional traffic, the jamming
signal informs all active stations about the collision.
2. After being informed about the collision, each station wishing to transmit
data must wait a random amount of time before initiating the whole process
again, starting with carrier sensing etc.
Three CSMA/CD variations provide increasing degrees of sophistication and
optimization:
o NON-PERSISTENT RULES
1. If the medium is idle, transmit.
2. If the medium is busy, wait a random amount of time and repeat step 1.
This method is very simple but inefficient, especially with high
utilization of the medium because the opportunity to transmit during a
brief idle period may be missed.
o 1-PERSISTENT RULES
1. If the medium is idle, transmit.
2. If the medium is busy, continue to listen until the medium is idle,
then transmit immediately.
3. If there is a collision, wait a random amount of time and repeat step
1.
If two or more stations are waiting to transmit, a collision will always
occur when the medium becomes available. However, since each station will
wait a random amount of time after a collision and before retransmitting,
things will sort themselves out from there. However the delays created by
the non-persistent rules for normal busy media will be avoided.
Within existing CSMA/CD implementations, this method is the most popular.
o P-PERSISTENT RULES
The p-persistent protocol tries to eliminate the collision that happens
with the 1-persistent protocol when the medium is busy and more than one
station is waiting to transmit. The rules are:
1. If the medium is idle, transmit with a probability of p. Hence there
is a probability of q=(1-p) that the station will wait for the next
time slot.
2. If the medium is busy, continue to listen until the medium is idle,
then repeat step 1.
3. If there is a collision, wait one time slot and repeat step 1.
The main problem with this approach is choosing a value for p. P must be
kept low to avoid instability, but this increases the wait probability
(q). This approach tends to be unstable with highly utilized media.
The 1-persistent protocol wastes less time after a collision than the
non-persistent protocol. The random backoff after a collision reduces the
chances of two stations contending for the medium again. Under high traffic
conditions, the 1-persistent rules are more like to run into collision and
thus create instability. However when p=1 then q=0, and therefore there is no
wait probability. Thus under normal or low traffic conditions the
1-persistent protocol can provide better throughput.
In conclusion, CSMA/CD or contention protocols in general perform best when
medium utilization is low. High utilization increases the number of
collisions and collision retransmissions, further adding to the utilization.
Sensitivity to increased traffic load is one of the major drawbacks of CSMA/CD
protocols.
2.2.1.1.1 Frame Format: Figure 9 under heading "2.2.1.1.1 Frame Format"
shows the frame format used for data transmission on a CSMA/CD LAN according
to the international standard specification issued by the IEEE 802.3. See
"3.0 LAN Architectures and Standards" . Note that this format is slightly
different from that used by "Ethernet" adapters.
+------------------------------------------------------------------------------
]
] Physical
] ]<------------- Physical Header -------------->] ]<-Trailer->]
] ] ]<------------------ FCS Protection ------------------->]
] *---------------------------------------------------------------------*
] ]Preamble] SD ] Destination ]Source ] LF ]Information] PAD ] FCS ]
] ] ] ] Address (DA)]Address (SA)] ] ] ] ]
] *---------------------------------------------------------------------*
] 7 1 ] ]2,6 ]2,6 2 ] 0-n 0-m] 4 ]
] ] Starting * ] ] Length Padding ]
] ] Delimiter ] ] Field ]
] ] B'10101011'] *-----* Frame
] ] ] ] Check
] Field ] ] Sequence
] Length *-------------------* I/G = Individual /Group address
] in Bytes ]I/G]15 address bits] B'0' B'1'
] *-------------------* U/L = Universally/Locally admin.
] <- 16 bit address --> B'0' B'1'
]
] *------------------------------------------------------*
] ]I/G]U/L]46 address bits ]
] *------------------------------------------------------*
] <------------------ 48 bit address field -------------->
]
]
+------------------------------------------------------------------------------
Figure 9. IEEE 802.3 MAC Frame Format
o Preamble- 7 padding bytes, allow the physical layer signalling (PLS)
circuitry to synchronize with the receive frame timing circuitry.
o SA, DA - 16-bit or 48-bit medium access control (MAC) address fields
according to the IEEE and ISO standards. A vendor is free to choose either
of them but all stations on a given LAN must use the same addressing
structure. Most current LAN implementations use 48-bit addressing, but
some early LANs use 16-bit address fields.
o LF - length field, indicating the actual length of the information field.
o Information field - may contain a LLC protocol data unit (LPDU).
o Padding - may contain padding characters if the minimum frame length
requirement is not met (512 bits).
o FCS - the result (32 bits) of a cyclic redundancy check algorithm
(specific polynomial executed against the contents of DA, SA, LF,
information and pad fields).
o An invalid frame is not passed to the next higher protocol layer (the LLC
sublayer), but is discarded by MAC level processing for any of the
following reasons:
- The frame does not contain an integral number of bytes.
- The content of frame length field is inconsistent with the actual
frame length.
- The FCS value calculated upon by the destination station does not
match the FCS value contained within the frame.
2.2.1.1.2 Additional CSMA/CD Considerations
o Cabling considerations
With CSMA/CD baseband bus topology, 50 ohm impedance coax wire is
recommended to reduce the capacitance caused by the taps. Almost all
other communications cables use an impedance of at least 75 ohms.
Frequently the long term implications of a particular cable type may be
overlooked during initial design and installation. If a CSMA/CD LAN using
50 ohm cable over time provides insufficient capacity for the required
data traffic, rewiring will usually be required and may turn out to be a
very costly operation.
o Probability of collisions
The probability of a collision occurring is proportional to the number of
stations, the number of transmissions per time interval, the length of the
LAN, the LAN utilization and the size and number of the collision windows
involved. A COLLISION WINDOW is the time during which a collision may
occur. Its maximum value is the time required by a signal to travel
between the two outmost stations on a CSMA/CD LAN. On a 10 Mbps CSMA/CD
baseband LAN the collision window size equals the one-way propagation
delay (sum of all transmission delays) between the contending stations.
While the size of collision windows may be decreased by reducing the frame
size used by adapters, it is not recommended since it will also increase
the number of windows and the overhead caused by frame headers and
trailers.
Baseband collision detection is usually accomplished by measuring the
signal strength on the bus. Therefore the signal between the two outmost
stations must be considerably stronger than normal noise levels.
Consequently, on a CSMA/CD baseband LAN, signal repeaters are generally
required every 500 meters. This has a positive consequence for the
recovery of relatively large frames. On the average, only about 512 bits
would have been transmitted when the collision is detected. This stops
the frame transmission and initiates the recovery process.
SLOT TIME is the minimum transmission time required by a station to ensure
it can hear all possible collisions and is equal to twice the maximum
collision window size. This implies a minimum frame size on a 10 Mbps
CSMA/CD baseband LAN of 512 bits. The frame header and trailer parts
require 208 bits, which leaves a 304-bit data field in a minimum length
frame. In shorter frames such as most control frames and short messages,
padding characters must be appended to the data field (refer to the frame
format). An increase in transmission speed increases the minimum frame
size.
Early CSMA/CD LANs had relatively few attachments such as processors and
terminal servers (controllers which served multiple fixed function
devices). With the advent of the intelligent programmable workstation,
the desire to attach such devices directly to the LAN may significantly
increase the number of attachments and alter traffic characteristics. Any
associated increase in collision probability should be considered in
designing a LAN which will use intelligent workstations.
o Collision resolution
As mentioned before, the first station to detect a collision immediately
transmits a JAMMING SIGNAL to notify all listening stations that a
collision has occurred. In response to this signal, each station enters a
BINARY EXPONENTIAL BACKOFF ALGORITHM, causing each station to wait for a
random amount of time before attempting to transmit again. If a station's
subsequent attempt results in another collision, its backoff delay is
doubled. This process may be repeated up to 16 times, after which the
station, if still unsuccessful, reports a transmission error. There is no
guaranteed delivery at the CSMA/CD MAC level. If traffic levels are high,
the more retries a station requires, the worse its chances to succeed may
be when attempting again.
o Basic performance
CSMA/CD protocols optimize throughput and response time under conditions
of relatively low (less than 40%) bandwidth utilization.
Test measurements indicate that typical user data frames contain about
1000 bits. Using this frame size to simulate the performance of a 10 Mbps
CSMA/CD LAN with 100 active LAN devices results in a maximum throughput of
about 3.6 Mbps because of the sensitivity of the protocol to load.
Increasing the transmission speed on a CSMA/CD LAN does not significantly
improve throughput since the minimum frame length increases
proportionally. It may also imply a requirement for more expensive media
and additional repeaters since signal attenuation is proportional to the
square of the transmission speed.
o LAN growth
In response to growth requirements, one might attempt to increase the
maximum allowable distance. However, this also implies increased slot
time, minimum frame length and larger collision windows for any given
load.
When such architectural constraints create a problem, they may be
addressed by bridging separate CSMA/CD LANs together to form an
interconnected CSMA/CD LAN as discussed earlier.
o CSMA/CD interconnection considerations
When attempting to project future load growth, it may be risky to
underestimate capacity requirements. Recent explosive growth in
communications bandwidth requirements is likely to continue in LAN
environments. Such applications as image and file servers will quickly
use the spare capacity now available.
More detailed analysis of future load growth will show the increasing
importance of LAN INTERCONNECTION and higher capacity BACKBONE LANS.
A potential problem for any LAN bridge is FRAME LOSS. Generally, frame
loss can occur because of high loads on the LAN or at the bridge level
(for example, a bridge running out of queuing capacity).
Excessive frame loss in a CSMA/CD bridge (resulting from collisions) is a
much more significant problem. CSMA/CD LANs have no priority mechanism at
the MAC level, and bridge adapters (for example, to backbone LAN segments)
by nature tend to experience the highest loads, and will be the most
likely ones to be involved in collision recovery. A bridge may also be
subject to peak load situations on both sides of the bridge, further
decreasing the probability of successful frame transmission across the
bridge.
The probability of successful end-to-end data transmission is obtained by
multiplying the probabilities associated with each bridge in the path
together. This may lead to unacceptable situations if the load increases
in one or more LAN segments. Unfortunately backbone LANs, only accessible
via bridges, may be the most likely ones to experience high traffic loads.
Thus care should be taken during design to ensure that such backbones have
relatively few attachments (that is, bridges).
A backbone LAN sometimes needs to offer a transmission speed exceeding the
speed of the lower level LANs that are interconnected. As explained
before, 10 Mbps is a practical limit for CSMA/CD LANs. Backbone LANs
based upon other medium access protocols may offer a more attractive
solution, despite the need for installation of different cabling for the
backbone.
A common routing technique used in CSMA/CD bridges is based on tables.
Tables are dynamically built at the bridge level containing station
addresses identified on both sides of the bridge. These tables may become
quite large in an expanding LAN environment, and may consume considerable
buffer space and processing capacity for table lookup and maintenance.
This must be considered when configuring such bridges. In CSMA/CD LANs
parallel active bridges to support heavier traffic would cause duplicate
frames and collisions and thus are prohibited. One bridge routing
algorithm, called spanning tree, does not allow duplicate frames and
provides mechanisms to ensure that parallel bridges or routes are not
concurrently active. Bridge routing techniques are discussed in "6.3
Bridge Standards."
2.2.2 Basic Token-Passing Ring Concepts
A token-ring network consists of the ATTACHING MEDIUM and RING STATIONS
(devices able to attach to the ring and to use the link access protocols).
A token-ring network uses one of several twisted pair media specifications,
each having its own price/performance ratio, and all suitable to carry most
other data communications signals. A token-ring network may also use optical
fiber media.
A single segment token-ring LAN installation is illustrated in Figure 10 under
heading "2.2.2 Basic Token-Passing Ring Concepts".
+------------------------------------------------------------------------------
]
] A.) Normal operation on Primary Path
] --------------------------------
] *--* *--* *--* *--*
] ]S9] ]S8] ]S7] ]S6]
] **** **** Primary Path **** ****
] *-<--* ]] ]] *--<----<----<---<* ]] ]] *----*
] ] *]-]]---]]------------]* *]-]]--------]]------]* ]
] ] ]*<**-<-**-<----<---<-*] ]*<**-<---<--**-<---<*] A
] ] ..]......................]...............].....................].. ]
] ] . ]RO PWC 4 RI] Backup Path ]RO PWC 3 RI] . ]
] ] . *----------------------* (Idle) *---------------------* . ]
] ] . . ]
] ] . *----------------------* *---------------------* . ]
] ] . ]RI PWC 1 RO] ]RI PWC 2 RO] . ]
] ] ..]......................]...............].....................].. ]
] ] ]*-**-->--->---->--->-*] ]*-**->---->---->-**-*] ]
] ] *]-]]--------**-------]* *]-]]---**--------]]-]* ]
] *-->-* ]] ]] *--->---->---->---* ]] ]] ]] *----*
] **** **** **** **** ****
] ]S1] ]S2] ]S3] ]S4] ]S5]
] *--* *--* *--* *--* *--*
]
]
] B.) Backup operation on Primary and Backup Path
] -------------------------------------------
] *--* *--* *--* *--*
] ]S9] ]S8] ]S7] ]S6]
] **** **** Primary Path **** ****
] *--<--*]] ]] *--<----<----<----<-*]] ]] *--<--*
] ] *>-*]]] ]] ]*-->---->---->---->]]] ]] ]*>-* ]
] ] ] *]]]]---]]-----------]]* Backup Path -]]**--------**-----]]- ] ]
] ] ] ]]***-<-**-<----<----*]] ]]*--------<--------*]] ] ]
] ] ] ]*->---->---->--->----*] ]*->---->---->---->--*] ] ]
] ] ] ]RO PWC 4 RI] ]RO PWC 3 RI] ] ]
] ] ] *----------------------* *---------------------* ] ]
] ] ] ] ]
] ] ] *----------------------* *---------------------* ] ]
] ] ] ]RI PWC 1 RO] ]RI PWC 2 RO] ] ]
] ] ] ]*---<----<---<----<--*] ............. ]*--<------<------<--*] ] ]
] ] ] ]]***->---->--->---->-*] ------------- ]*-**->---->--->--***]] ] ]
] ] ] *]]]]--------**--------* Broken Cable *--]]---**--------]]]]* ] ]
] ] *<-*]]] ]] Segment ]] ]] ]]]*-<* ]
] *->---*]] ]] (removed) ]] ]] ]]*--->-*
] **** **** **** **** ****
] ]S1] ]S2] ]S3] ]S4] ]S5]
] *--* *--* *--* *--* *--*
]
]
] RI = Ring In PWC = Passive Wiring Concentrator
] RO = Ring Out Sn = LAN Station
]
+------------------------------------------------------------------------------
Figure 10. Token-Passing Ring Data Paths
This figure shows four PASSIVE WIRING CONCENTRATORS (PWC) and 9 physically
attached nodes. The power from the attached nodes when transmitted to the
concentrator activates relays in the concentrator to allow the station to send
signals across the LAN to other stations. In Part A of Figure 10 under
heading "2.2.2 Basic Token-Passing Ring Concepts" two adapters (nodes S2 and
S4) have not powered their respective PWC relays and therefore their lobe
wires are internally bypassed. In Part B of Figure 10 under heading "2.2.2
Basic Token-Passing Ring Concepts" four adapters (nodes S2, S4, S6 and S7) are
not actively inserted into the ring (their lobe wires are internally bypassed)
and the primary path is wrapped to the backup path in PWCs 1 and 2.
When a cable segment between PWCs fails, manual removal from the appropriate
ring-in and ring-out connectors causes automatic wrapping of the primary path
to the backup path. How such a permanent wire fault is reported for LAN
management is explained in the discussion of beaconing later on in this
section. Recovery from the same error is automatic when using the IBM 8230
Controlled Access Unit, which is an Active, or powered, Wiring Concentrator
(see "5.1.1.2 IBM 8230 Controlled Access Unit")
A ring station transfers data to the ring, in a data transmission unit called
a FRAME. Frames are sent sequentially from one station to the next station
physically active on the ring. This station is called the DOWNSTREAM
NEIGHBOR. Each ring station repeats the frame. While doing so it performs
error checking on the bit stream and it will copy the data if its own address,
either its MAC or any of its Functional Addresses, are identified as a
destination station in the frame. Upon return of the frame to the originating
station, the latter will remove the data from the ring.
In a token-passing protocol, a ring station can only transfer data to the ring
while it is holding a TOKEN. The token is a specific bit sequence (24 bits)
circulating around the ring at a rated speed (4 Mbps or 16 Mbps in current
implementations, 100 Mbps according to the Fiber Distributed Data Interface
specification; see "2.3 Fiber Distributed Data Interface Concepts"). Because
of the high transmission speed with respect to the total ring length, a short
ring might contain only a few bits at any point in time. Only one token may
exist on a ring segment at any given point in time. Therefore, a delay
equivalent to the time for a token to circulate the ring is required to ensure
that no overrun occurs which would result in a station receiving a token that
it is transmitting and thinking that a second token exists on the ring. For a
24-bit token this means a minimum 24-bit delay. In addition to this delay an
additional elastic buffer is introduced to support the token protocols and
speed.
In order to establish communication between any two ring stations, ADDRESSING
mechanisms are needed. At the same time the integrity of the transmitted
frames between ring stations must be preserved. Therefore data checking
capabilities are required at the MEDIUM ACCESS CONTROL level of a ring
station.
2.2.2.1 MAC Addressing
Any ring station is identified by a unique INDIVIDUAL ADDRESS. This address
can be UNIVERSALLY ADMINISTERED, assigned by the IEEE organization (see "3.0
LAN Architectures and Standards"). Because it is set in read-only memory
(ROM) on a token-ring adapter card, the universally administered address is
also called a BURNED-IN ADDRESS.
Some manufacturers have been assigned universal addresses that contain an
organizationally unique identifier. For instance, IBM has an identifier of
x'10005A'. All IBM token-ring cards that use IBM token-ring chip sets, have
the first 6 digits of their address begin with those characters. Other
identifiers are x'000143' for IEEE 802, and x'1000D4' for DEC.
IEEE universal addresses, whether for token-ring or 802.3 stations are all
allocated out of the same common pool, but uniqueness is guaranteed.
A ring station's individual address can also be LOCALLY ADMINISTERED, that is
set at adapter-open time and typically defined by a network administrator.
A number of destination ring stations can be identified by a GROUP MAC
ADDRESS. Some standard group addresses have been defined. These are listed in
Figure 11 under heading "2.2.2.1 MAC Addressing".
+------------------------------------------------------------------------------
]
]
] *--------------------------------------------------------*
] ] Bridge ] X'8001 4300 0000' ]
] ]------------------------------------+-------------------]
] ] Bridge management ] X'8001 4300 0008' ]
] ]------------------------------------+-------------------]
] ] Load Server ] X'8001 4300 0088' ]
] ]------------------------------------+-------------------]
] ] Loadable device ] X'8001 4300 0048' ]
] ]------------------------------------+-------------------]
] ] ISO 10589 level-1 Intermediate Stns] X'8001 4300 0028' ]
] ]------------------------------------+-------------------]
] ] ISO 10589 level-2 Intermediate Stns] X'8001 4300 00A8' ]
] ]------------------------------------+-------------------]
] ] FDDI RMT Directed Beacon ] X'8001 4300 8000' ]
] ]------------------------------------+-------------------]
] ] FDDI status report frame ] X'8001 4300 8008' ]
] ]------------------------------------+-------------------]
] ] OSI Network Layer End-stations ] X'9000 D400 00A0' ]
] ]------------------------------------+-------------------]
] ] OSI NL Intermediate stations ] X'9000 D400 0020' ]
] *--------------------------------------------------------*
+------------------------------------------------------------------------------
Figure 11. Standardized Group Addresses
A token-ring LAN also provides a special case of a locally administered group
address called FUNCTIONAL ADDRESSES. Each (bit-significant) functional
address represents a well-identified server function within the access
protocol. Of 31 possible functional addresses, 14 have been defined while the
remaining ones are reserved for future use or may be user-defined. They are
listed in Figure 12 under heading "2.2.2.1 MAC Addressing".
+------------------------------------------------------------------------------
]
] *-------------------------------------------------*
] ] Active Monitor ] X'C000 0000 0001' ]
] ]-----------------------------+-------------------]
] ] Ring Parameter Server ] X'C000 0000 0002' ]
] ]-----------------------------+-------------------]
] ] Network Server Heartbeat ] X'C000 0000 0004' ]
] ]-----------------------------+-------------------]
] ] Ring Error Monitor ] X'C000 0000 0008' ]
] ]-----------------------------+-------------------]
] ] Configuration Report Server ] X'C000 0000 0010' ]
] ]-----------------------------+-------------------]
] ] Synchronous Bandwidth Mgr. ] X'C000 0000 0020' ]
] ]-----------------------------+-------------------]
] ] Locate - Directory Server ] X'C000 0000 0040' ]
] ]-----------------------------+-------------------]
] ] NETBIOS ] X'C000 0000 0080' ]
] ]-----------------------------+-------------------]
] ] Bridge ] X'C000 0000 0100' ]
] ]-----------------------------+-------------------]
] ] IMPL Server ] X'C000 0000 0200' ]
] ]-----------------------------+-------------------]
] ] Ring Authorization Server ] X'C000 0000 0400' ]
] ]-----------------------------+-------------------]
] ] LAN Gateway ] X'C000 0000 0800' ]
] ]-----------------------------+-------------------]
] ] Ring Wiring Concentrator ] X'C000 0000 1000' ]
] ]-----------------------------+-------------------]
] ] LAN Manager ] X'C000 0000 2000' ]
] ]-----------------------------+-------------------]
] ] ] X'C000 0000 8000' ]
] ] User-defined ] through ]
] ] ] X'C000 4000 0000' ]
] *-------------------------------------------------*
]
]
+------------------------------------------------------------------------------
Figure 12. IEEE and IBM Functional Addresses
The most relevant protocol server functions will be described in greater
detail in the bridge and LAN management sections (see "6.6 IBM LAN Bridges"
and "9.2 LAN Management and MAC Architecture").
In addition two special destination address values have been defined. The
ALL-STATIONS BROADCAST group address X'FFFFFFFFFFFF' identifies all ring
stations as destination stations. A frame carrying the individual NULL
ADDRESS X'000000000000' as its destination MAC address is not addressed to any
ring station, therefore it can only be sent but not received.
IEEE allows vendors to implement either 16-bit or 48-bit MAC addresses. The
actual address field formats are shown in the picture Figure 13 under heading
"2.2.2.1 MAC Addressing".
+------------------------------------------------------------------------------
]
]
] *------------------------------------------------------*
] ]I/G]U/L]46 address bits ]
] *------------------------------------------------------*
] <------------------ 48 bit address field -------------->
]
] *-------------------* I/G = Individual /Group address
] ]I/G]15 address bits] B'0' B'1'
] *-------------------* U/L = Universally/Locally administered
] <- 16 bit address --> B'0' B'1'
]
]
]
+------------------------------------------------------------------------------
Figure 13. IEEE LANs - MAC Address Format
For the IBM LAN implementations, 48-bit addressing has been selected. The
implementation format (8 under heading "2.2.2.1 MAC Addressing") is shown in
Figure 14 under heading "2.2.2.1 MAC Addressing".
---Footnote---
(8) The universal/local address field (Byte 0, Bit 1) in a source address is
used to indicate the presence of routing information; see "6.6 IBM LAN Bridges.
--------------
+------------------------------------------------------------------------------
]
]
] 0 1 2 3 4 5
] *--------------------------------------------------------*
] ]I]U] ] ]F] ] ] ] ]
] ]/]/]<-- reserved ->]A] ] ] ] ]
] ]G]L] ] ]I] ] ] ] ]
] *--------------------------------------------------------*
] 0 1 234567 01234567 0 1234567 01234567 01234567 01234567
] <------------ 31 bits ----------->
] <------------------ 48 bit address -------------------->
]
] FAI = Functional Address Indicator
] (only significant if Byte 0, bits 01 = B'11')
]
]
+------------------------------------------------------------------------------
Figure 14. IBM Token-Ring Network - MAC Address Format
o The reserved bits are set to B'0' for locally administered addresses.
o Functional address indicator = B'0' indicates a functional address if I/G
= B'1' (indicating a group address).
o For individual locally administered addresses, FAI must be B'0' by
convention. This is an addressing anomaly.
These rules yield valid address ranges as described in Figure 15 under heading
"2.2.2.1 MAC Addressing" for any IBM Token-Ring Network adapter.
+------------------------------------------------------------------------------
]
]
] *------------------------------------------*
] ] I/G ] U/L ] FAI ] Definition/Range ]
] *------------------+-----+-----+-----+------------------------]
] ] Individual, ] 0 ] 0 ] 0/1 ] Mfg_Code,Serial_Number ]
] ] Universally Adm. ] ] ] ] (IEEE assigned) ]
] ]------------------+-----+-----+-----+------------------------]
] ] Individual, ] 0 ] 1 ] 0 ] X'4000 0000 0000' - ]
] ] Locally Admin. ] ] ] ] X'4000 7FFF FFFF' ]
] ]------------------+-----+-----+-----+------------------------]
] ] Group ] 1 ] 1 ] 1 ] X'C000 8000 0000' - ]
] ] Address ] ] ] ] X'C000 FFFF FFFF' ]
] ]------------------+-----+-----+-----+------------------------]
] ] Functional ] 1 ] 1 ] 0 ] X'C000 0000 0001' - ]
] ] Address ] ] ] ] X'C000 0000 2000' ]
] ] ] ] ] ] (bit-sensitive) ]
] *-------------------------------------------------------------*
]
] Functional Addresses are listed in Figure 12 under heading "2.2.2.1
]
]
+------------------------------------------------------------------------------
Figure 15. Valid Address Ranges
2.2.2.1.1 Data Transmission: The transmission technique used in token
passing ring LANs is BASEBAND TRANSMISSION.
In a token-ring LAN, high-order bytes/bits are transmitted first, that is byte
0 is transmitted before byte 1 and high-order bit 0 within a byte (of 8 bits)
is transmitted first. This transmission order can be different for other types
of LAN segments using a different access protocol, for example, CSMA/CD.
Opposite transmission order may be a diagnostic consideration when evaluating
trace information from LAN segments of different nature because of the
possible need to reorder the bits. The ability to reorder the bits without
significant performance degradation, may also be a functional requirement of
bridge products being considered for LAN segment interconnection.
Figure 16 under heading "2.2.2.1.1 Data Transmission" shows the format of the
information to be transmitted on a token passing-ring.
+------------------------------------------------------------------------------
]
]
] * Frame Format Physical
] Trailer
] ]<---------- Physical Header ------------>] ]<--->]
] ] ]<--------- Code Violation Protection-------------->]
] ] ] ]<-------------- FCS Protection ----------------->]
] ] ] ] ]< Information Field >] ]
] *---------------------------------------------------------*
] ]S]A]F]Destination]Source ]Routing ] ]F]E]F]
] ]D]C]C]Address ]Address ]Information] D A T A ]C]D]S]
] ] ] ] ] (DA) ] (SA) ] (RI) ] ]S] ] ]
] *---------------------------------------------------------*
] ] ] ] ] ] ] ] ]
] ] ] Frame Control ] ] ] ] Frame Status
] ] ] FF]ZZZZZZ ] ] ] ] A]C]rr]A]C]rr
] ] ] ] ] ] Ending Delimiter
] ] ] ] ] ] VV1VV1]I]E
] ] Access Control ] ] Frame Check Sequence
] ] PPP]T]M]RRR ] ] (4 byte CRC value)
] ] ] LLC protocol data unit
] ] ] or MAC management information
] ] ]
] Starting Delimiter interconnected segments, LLC only,
] VV0VV000 present when RI Indicator = B'1'
] (Byte 0, bit 0 of SA)
]
] * Token format *--------------*
] ] SD ] AC ] ED ]
] *--------------*
]
] * Abort Delimiter *---------*
] Format ] SD ] ED ]
] *---------*
]
]
]
+------------------------------------------------------------------------------
Figure 16. IEEE 802.5 MAC Frame and Token Format
Explanations for abbreviations used in the above figure include:
o SD - starting delimiter consisting of eight bits "VV0VV0 00" and
ED - ending delimiter consisting of eight bits "VV1VV1 I E" where:
- VV1VV1 refers to a '1' bit code violation (J) followed by a '0' bit
violation (K). This is a function of the Differential Manchester
encoding scheme used in the Token-Ring Network. Refer to IBM
TOKEN-RING NETWORK ARCHITECTURE REFERENCE for a description of the use
of Differential Manchester code violations to define starting and
ending delimiters.
- I is an intermediate bit, B'0' which indicates a single-frame or the
last frame of a multiple-frame transmission using a single token.
- E is an error-detected bit. B'1' indicates existence of a code
violation between the starting and ending delimiters, a non-integral
number of bytes in the frame or a cyclic redundancy check (CRC) error.
o AC - access control consisting of eight bits "PPP T M RRR" where:
- PPP are priority bits (low B'000' to high B'111'). A ring station can
use a token at a priority less than or equal to the ring station's
allowed access priority.
- RRR are reservation bits (low B'000' to high B'111') which allow a
station with high access priority to request that a subsequent token
be issued at the required priority.
- T is the token bit (B'0' = token, B'1' = frame)
- M is a monitor bit used to prevent continuously circulating frames or
non-zero priority tokens. Refer to the functions of the active
monitor in the discussion of "2.2.2.3 Token Monitoring" for more
details.
o FC - frame control consisting of eight bits "FF ZZZZZZ" where:
- FF are Frame Type bits (B'00' = MAC frame, B'01' = LLC frame).
- ZZZZZZ are Control bits (MAC buffering information or 000YYY where YYY
indicates the intended LLC sublayer message priority).
o FS - frame status is an eight-bit field "A C rr A C rr" where:
- A is an address recognized bit (B'1' indicates that a valid station
has recognized the destination address)
- C is a frame copied bit
- rr are reserved bits
These bits are duplicated within the field because they not covered by
the frame check sequence protection.
o Any originating ring station can abort a frame that is being transmitted
at any time by transmitting an ABORT DELIMITER.
The information field in the MAC frame format has been split up into an
optional routing information field and data. See "6.3.2 Source Routing." In
an LLC frame, the data field will contain a LOGICAL LINK CONTROL PROTOCOL DATA
UNIT (LPDU, see "3.2.2.3 LLC Protocol Data Unit").
When the frame control field indicates a MAC frame (because bits 0 and 1 of
the field are equal to B'00') the data field contains specific MAC control
information. It consists of a major vector length (LL), a major vector
identifier (MVID) and zero or more subvectors as shown in Figure 17 under
heading "2.2.2.1.1 Data Transmission". The command field within the MVID
contains bit values (called code points) that uniquely identify the type of
MAC frame.
+------------------------------------------------------------------------------
]
]
] *------------------------------------/ /---------------*
] ] MAC ] MAC ] MAC ] ] MAC ]
] ] LLID ] Subvector-0 ] Subvector-1 ] ] Subvector-n ]
] *------------------------------------/ /---------------*
] ] ] ] ]
] *------* *----------------------* *--------* *--------*
] ]01234567 01234567 0123 4567 01234567] ]01234567 01234567 min. 2 bytes]
] *------------------------------------* *-------------------------------*
] ] Major Vector ] DC ] SC ]Command ] ] Subvector ] Value ]
] ] Length (LL) ] ] ] ] ] Length ] Id ] ]
] *------------------------------------* *-------------------------------*
] <------ MVID ------> ] ]
] Major Vector Id ] *----------*
] ] 0 7]
] C/S Common/Specific Indicator *-------------------*
] R/O Required/Optional Indicator ]C/S]R/O]P]P]P]P]P]P]
] P Code Point bit *-------------------*
]
]
+------------------------------------------------------------------------------
Figure 17. Token Passing Ring MAC Frame - Data Field Format
Examples of MVID code points are X'05' to indicate an Active Monitor Present
MAC frame, X'02' for a Beacon MAC frame, etc. Unique MVID values for 28
different MAC frames have been defined. A complete description can be found in
the IBM TOKEN-RING NETWORK ARCHITECTURE REFERENCE. Subvectors provide
additional information depending on the specific major vector identifier.
MAC Frames are processed according to destination and source function classes,
including:
o Ring Station
The functions necessary for connecting to the LAN and for operating with
the token-ring protocols. A ring station is an instance of a MAC sublayer
in a node attached to a ring.
o DLC_LAN_MGR
The manager function of the data link control component activates and
deactivates ring stations and link stations on command from the physical
device. It also manages information transfer between data link control and
the physical device.
o Configuration Report Server (CRS)
The CRS function can reside on each ring in a multi-segment environment in
which configuration is being monitored. This function receives
notifications about station insertion and removal, and notifications about
active monitor failures.
o Ring Parameter Server (RPS)
The RPS function can reside on every segment in a multi-segment ring
environment on which operational parameters are centrally managed. It may
provide operational values to attaching stations upon request. For
example, a ring station may request such parameters as ring number upon
insertion into the ring.
o Ring Error Monitor (REM)
The REM server function is present on segments for which errors are to be
monitored or analyzed. It collects error information from LAN stations
attached to the local ring, analyzes soft error reports and may forward
error reports to a LAN Manager.
o RPL server
The RPL server function and its RPL functional address are involved during
the power-on process of a LAN station equipped with the remote program
load feature. Such a station will insert into the ring to find a control
program server on the ring from which to download its control program and
complete its initialization processes.
2.2.2.1.2 The Token Passing Ring Protocol: To transmit data on the LAN
medium, a ring station captures a TOKEN, and sets the token bit in the access
control field to identify that the data being transmitted is a frame. To this
header the transmitting station appends destination and source MAC addresses,
data, a newly calculated frame check sequence field, and the ending delimiter
and frame status fields.
Any subsequent station will receive and retransmit the frame while performing
a CRC check. Such a station is said to be in NORMAL REPEAT MODE.
In general, a ring station in normal repeat mode checks the data in the tokens
and frames it receives and sets the error-detected, address recognized or
frame copied bits in a frame (bits E, A, or C) as appropriate while repeating
the signal. A destination station will copy the data (FRAME COPYING) and pass
the frame on. While processing the frame trailer, the destination station
marks the A and C bits. Upon return to the originating ring station, the frame
is removed from the ring and the A and C bits in the frame trailer's FS field
are checked to see if the frame was recognized and read by the destination
station or a bridge (this occurs for MAC frames only). When the frame header
is received by the originating station the originating ring station must
release a new token, possibly at a different priority level, for another ring
station to capture and proceed with data transmission. The priority
reservation bits in the access control field of the returned frame together
with stored priority levels in the originating station determine the priority
of the new token. See the description of ACCESS PRIORITY later in this
section for details.
This protocol is called a SINGLE TOKEN protocol, since only one token can
circulate on the ring at any time.
2.2.2.2 Early Token Release
If an originating station releases a new token only when the frame header has
circulated around the ring back to the source and the frame transmission time
is shorter than the ring transmit time, then the originating station must
generate idles until a header is received.
Token-passing ring LAN protocols define the length of a token to be 24 bits
and the shortest possible MAC frame to be 200 bits long. On a 4 Mbps
token-ring LAN where the length of 1 bit is roughly 50 meters, a complete
token is 1,200 meters long while the shortest frame length would be 10,000
meters. Therefore at 4 Mbps, the percentage of potential bandwidth which
remains idle can be extremely small (that is high bandwidth utilization can be
maintained at higher traffic levels.)
If, however, we consider a 16 Mbps token-ring LAN, 1 bit being 12.5 meters
long, a complete token and the shortest possible MAC frame become four times
smaller (300 and 2,500 meters respectively). Now we may wish to optimize the
utilization of the medium by reducing the idle time required by waiting for a
header. Obviously when moving to even higher transmission speeds, for
example, a 100 Mbps FDDI LAN the token passing protocol must be adjusted to
achieve better utilization of the potential bandwidth.
The architecture provides an option called EARLY TOKEN RELEASE. With this
option a transmitting station will release the token after completing
transmission of the data frame before receipt of the header of the transmitted
frame, thereby eliminating the idle time while waiting for the header to
reappear. When such early release has occurred, an adapter indicator is set
to prevent the adapter from releasing another token upon return of the frame
header. This allows multiple frames, but still only ONE token to coexist on
the LAN.
The early token release option is enabled by default on the 16 MBPS IBM
TOKEN-RING NETWORK. It is an option for each station and it is not required
that all stations implement the option.
2.2.2.3 Token Monitoring
Token-passing protocols provide relatively greater control and management at
the medium access control (MAC) level than that provided by CSMA/CD protocols.
The token passing ring protocol concepts, described in the following sections,
are implemented in the adapters themselves. They contribute to the
availability, performance and manageability of a token-ring LAN.
At any point in time, one station and only one per segment performs an ACTIVE
MONITOR function. Any ring station can act as the active monitor. Only one
however, will have this function enabled.
Active monitor tasks support monitoring of the token and other ring management
functions such as:
o Detection and recovery of a lost token or frame, including initiation of a
token when a ring is started.
o Detection and recovery of a circulating priority token or frame.
o Detection and recovery of multiple tokens on the ring.
o Detection and recovery of multiple active monitors.
o Timing control to ensure accurate transmission regardless of the ring
length.
All other ring stations are said to be STANDBY MONITORS, prepared to take over
the active monitor function should it fail.
The following description summarizes how the active monitor performs its ring
management tasks:
o In every transmitted token or frame, the MONITOR BIT (M) in the access
control field is initially set to B'0'. As the active monitor repeats a
frame or non-zero priority token, the M-bit is set to B'1'. If the M-bit
had already been set to B'1', the active monitor assumes that the frame or
token has already circled the ring once and that the originating station
has not been able to remove the frame or priority token. The active
monitor will purge the ring and generate a new token.
o To ensure that a complete (24-bit) token can be transmitted before the
token returns to the originating ring station, the active monitor
introduces a 24-bit ring delay.
o The active monitor periodically broadcasts an ACTIVE MONITOR PRESENT MAC
frame. This forces each station on its ring to acquire the address of its
NEAREST ACTIVE UPSTREAM NEIGHBOR (NAUN) and to initiate a number of
control timers within each station. This information is used when
isolating errors on a segment.
o Loss of a token or frame is detected by expiration of an ANY-TOKEN TIMER
whose time-out value exceeds the time required for the longest possible
frame to circle the ring. The active monitor restarts this timer each time
it transmits a starting delimiter. Upon expiration of this timer, the
active monitor assumes a lost token or frame, purges the ring and
originates a new token. The any-token timer value is defined as the sum
of the physical trailer transmission delay plus the delay to transmit the
longest frame. The IEEE 802.5 name for this timer is
Valid_Transmission_Timer.
2.2.2.3.1 Ring Purge: To purge the ring, the active monitor broadcasts a
RING PURGE MAC frame (indicated by X'04' in the frame control field) before
originating a new token. Return of the Ring Purge MAC frame indicates proper
signal propagation around the ring. The Ring Purge frame resets the ring
stations to normal repeat mode, canceling or restarting all the appropriate
timers. The active monitor starts a Ring-Purge Timer when sending the purge
frame. This timer will expire if the frame can't circulate and the monitor
will enter a recovery process called CLAIM TOKEN.
2.2.2.4 Neighbor Notification
The NEIGHBOR NOTIFICATION PROCESS begins when the active monitor transmits an
Active Monitor Present MAC frame to all stations on the ring (single ring
broadcast). The first ring station that receives the Active Monitor Present
MAC frame copies it (if possible) and sets the address-recognized (A) and
frame-copied (C) bits to B'1'. It then saves the source address from the
copied frame as its NAUN address (the address of the active monitor) and
starts a timer called the Notification-Response timer.
All other active stations on the ring repeat, but do not otherwise process the
Active Monitor Present MAC frame because the frame's A and C bits have already
been set.
When the Notification-Response timer of the first station downstream from the
active monitor expires, it broadcasts a Standby Monitor Present MAC frame.
The next ring station downstream copies its NAUN address from the source
address field of the Standby Monitor Present frame, sets the A and C bits to
B'1', and starts its own Notification-Response timer. When this timer
expires, this station in turn transmits its Standby Monitor Present MAC frame.
In this way, Neighbor Notification proceeds around the ring, with each ring
station receiving and transmitting Standby Monitor Present MAC frames until
the active monitor copies its NAUN address from a Standby Monitor Present MAC
frame. The active monitor then sets the Neighbor-Notification Complete flag
to B'1', indicating that the process has been successfully completed.
Neighbor-Notification thus enables a ring station to learn its NAUN address,
and to provide its address to its downstream neighbor.
2.2.2.4.1 Standby Monitor: Any ring station that is not performing the
active monitor function acts as a STANDBY MONITOR. Its purpose is to detect a
failing active monitor and disruptions on the ring.
Each time a token or frame is repeated, a standby monitor restarts its
GOOD-TOKEN TIMER ( > Any-token timer) to verify the presence of an active
monitor.
A second timer, the RECEIVE-NOTIFICATION TIMER, is restarted by a standby
monitor each time it copies an ACTIVE MONITOR PRESENT MAC frame.
If any of these two timers expires, the standby monitor station will initiate
the token-claiming process.
2.2.2.4.2 The Token-Claiming Process: This process, also called the
Monitor-Contention Process, is the procedure by which ring stations elect a
new active monitor. This process is started upon any of the following
conditions by:
o The active monitor detecting
- Loss of signal
- The Active Monitor Present MAC frame doesn't return
(Receive-Notification Timer expires)
- Failure of Ring-Purge MAC frames to return completely (Ring Purge
Timer expires)
o A standby monitor detecting
- Loss of signal
- Absence of active monitor's token management functions (good_token
timer expires)
- Missing Neighbor_Notification process (Receive_Notification_Timer
expires)
o A ring station attaches to the ring and does not detect an active monitor
(for example, when it is the first station on the ring).
The ring station detecting one of these conditions enters Claim-Token-Transmit
mode by broadcasting a CLAIM TOKEN MAC frame and repeating it at a defined
interval. Each participating ring station (9 under heading "2.2.2.4.2 The
Token-Claiming Process")
---Footnote---
(9) Every ring station contains an option that indicates whether it will
participate in token claiming. The default is not to participate. However, if
the ring station detects the condition, it has to initiate the process.
--------------
compares the address in the Claim Token MAC frame's source address field to
its own.
o If the source address is greater than the ring station's address, the
station enters Claim Token Repeat operating mode.
o If the source address is less than the ring station's address, the station
transmits its own Claim Token MAC frames.
o If the source address is the same as the ring station's address, it
continues broadcasting until it has received three of its own Claim Token
MAC frames. This indicates that the ring is viable and the station has
won token-claiming.
The station then adds the token delay to the ring, purges the ring, starts its
active monitor timers, and releases a new token. It is now the new active
monitor.
2.2.2.4.3 Ring Station Insertion: This process is executed by any ring
station when entering the ring. It is also known as the five-phase insertion
process:
o Phase 0: Lobe testing.
A series of Lobe Media Test MAC frames is sent on the lobe wire to the
multi-station access unit. The signal is wrapped at the entry into the
multistation access unit causing the frames to return to the station.
Then the receive logic is tested. If the tests are successful, a five-volt
DC voltage (also called PHANTOM CURRENT) is sent to open the relay and
attach to the ring.
o Phase 1: Monitor check.
The attaching station starts its Insert timer, and watches for an
Active_Monitor_Present, Standby_Monitor_Present or Ring Purge MAC frame
before this timer expires. If the timer expires, token-claiming is
initiated. When "first station on ring", the attaching station will
become the active monitor.
o Phase 2: Duplicate Address check.
The station sends a Duplicate Address Test MAC Frame (destination address
= source address = station's unique address). If a duplicate address is
found (A-bit = B'1'), the station detaches from the ring.
o Phase 3: Participation in Neighbor_Notification
The station learns its nearest active upstream neighbor (NAUN) and reports
its own address to its nearest active downstream neighbor.
o Phase 4: Request Initialization
A Request Initialization MAC frame is sent to the ring parameter server,
if present (if not, default values will be used). The ring parameter
server responds with a Initialize_Ring_Station MAC frame. Parameters
which can be set are physical location, soft error report timer value,
ring number and ring authorization level. In this way the last three
parameters may be set to the same values for all stations on the ring.
2.2.2.4.4 Hard-Error Detection and Reporting: A HARD ERROR is a permanent
fault that stops normal traffic on the ring. It is usually detected first at
the receive side of the ring station downstream from the fault. A change in
ring configuration is required to bypass such a fault and to restore normal
operation. Reconfiguration may be the result of automatic recovery or, if this
process fails to bypass the error, it may require manual intervention.
When a ring station detects a hard error, it starts transmitting at a
specified time interval BEACON MAC frames until its input signal is restored
or until it removes itself from the ring. The detecting station also starts a
Beacon timer. All other stations enter Beacon_Repeat mode when they receive a
Beacon MAC frame.
A beacon frame identifies the address of the nearest active upstream neighbor
of the beaconing station as well as error type information.
When the beaconing station's NAUN has copied a number of these beacon frames,
the NAUN will go offline and perform microcode and lobe tests. If the tests
are successful, the station reattaches to the ring immediately. If the tests
fail, the station stays offline.
When the beacon timer expires in the detecting (beaconing) station, and normal
traffic has not been restored, the station assumes that its NAUN went offline,
found no errors and came back online. It will now go through the same process
as its NAUN. If the tests fail, the beaconing station remains detached. If
successful, the station reattaches immediately. In the latter case, normal
traffic may not have been restored during automatic recovery. Network
management will be informed (ref. "9.2 LAN Management and MAC Architecture")
and manual intervention will be required. While reporting a permanent hard
error, a set of adapter addresses is provided to identify the faulty part of
the ring as a small FAULT DOMAIN.
2.2.2.4.5 Soft-Error Detection and Reporting: Intermittent faults that
temporarily disrupt normal operation of the ring are called SOFT ERRORS. They
are usually tolerated by error recovery procedures but they may impair normal
ring operation if excessive or non-random.
The most critical soft errors are monitored in each ring station by means of a
set of counters. Every two seconds the values of the soft error counters are
sent as a Soft Error Report MAC frame to the Ring Error Monitor functional
address (typically residing in a bridge or LAN manager station), where the
values for each counter are accumulated. If a soft error counter exceeds a
predefined threshold, a LAN Manager will be informed through its link with the
LAN reporting mechanism. The LAN Manager may reconfigure the ring to bypass a
faulty node, if the fault can be located.
Soft errors are said to be ISOLATING if a fault domain can be specified. If
not, they are called NON-ISOLATING soft errors.
2.2.2.4.6 Access Priority: The following discussion applies both to 4 Mbps
and 16 Mbps token-ring LANs and is an integral part of the token passing ring
protocol. This access priority architecture is not applicable to the FDDI
protocol where access priority is based upon timers rather than the contents
of an access control field.
As stated earlier, access priority in a token or frame is indicated the first
three bits (PPP) of the access control field (AC). Any reservation of a
priority level is indicated in the last three bits (RRR) of the AC field by a
station requiring higher transmission priority.
A ring station wishing to transmit a frame at a given priority can use any
available token with a priority level equal to or less than the priority of
the frame to be transmitted. If such a matching token is not available, the
ring station may reserve a token of the required priority in a passing token
or frame according to the following rules:
o If the passing token or frame already contains a priority reservation
higher than the desired one, the ring station must leave the RRR bits
unchanged.
o If the RRR bits have not yet been set (RRR = B'000'), or indicate a lower
priority than the desired one, the ring station will set the reservation
bits to its required priority.
Upon removal of a frame by its originating station, the reservation bits in
the header are checked. If they show a non-zero value, the station must
release a non-zero priority token. The actual priority of the new token is
based on the priority used by the ring station for the recently transmitted
frame, the reservation received upon return of the frame and any stored
priority.
A ring station originating a token of higher priority enters PRIORITY-HOLD
STATE, (also called a STACKING STATION in the IEEE 802.5 token passing ring
standards).
Figure 18 under heading "2.2.2.4.6 Access Priority" lists the priority
definitions as provided by the IBM Token-Ring Network architecture.
This protocol option however, impacts the priority handling mechanism, since a
new token may be transmitted by the originating station before it is able to
verify the access control field in its returned frame.
If the frame header was already received, the token will be issued according
to the priority and reservation requested in the AC field of the frame and the
resulting priority levels stored in the station.
If the frame header has not yet been completely received by the originating
station, the token will be released with the same priority and reserved
priority as the transmitted frame.
+------------------------------------------------------------------------------
]
]
]
] *-----------------------------------------*
] ] B'000' ] Normal User Priority ]
] ] ] MAC frames that need no token ]
] ] ] Response type MAC frames ]
] ]--------+--------------------------------]
] ] B'001' ] Normal User Priority ]
] ]--------+--------------------------------]
] ] B'010' ] Normal User Priority ]
] ]--------+--------------------------------]
] ] B'011' ] Normal User Priority ]
] ] ] MAC Frames that need token ]
] ]--------+--------------------------------]
] ] B'100' ] Bridge ]
] ]--------+--------------------------------]
] ] B'101' ] reserved for IBM ]
] ]--------+--------------------------------]
] ] B'110' ] reserved for IBM ]
] ]--------+--------------------------------]
] ] B'111' ] Specialized Station Management ]
] *-----------------------------------------*
]
+------------------------------------------------------------------------------
Figure 18. Token Passing Ring Protocol - Priority Allocation Table
To prevent a high-priority station from monopolizing the LAN medium and to
make sure the priority eventually can come down again, the protocol provides
FAIRNESS within each priority level. Figure 19 under heading "2.2.2.4.6
Access Priority" shows how fairness within a priority is maintained.
-------------------------------------------------------------------------------
0*-* 1*-* 0*-* 1*-*
*----]A]--------]B]----* *----]A]--------]B]----*
] *-* *-* ] ] *-* *-* ]
*-* *-------> *-* *-* *-* *-*
]F] 0,0 ]C] ]F] 0,1] ]C]
*-* *-* *-* <-* *-*
] *-* *-* ] ] *-* *-* ]
*----]E]--------]D]----* *----]E]--------]D]----*
*-* 3*-* *-* 3*-*
1) Station A generates a frame with 2) Station B reserves a prio
P,R = 0,0 using a non-priority token. of 1 (P,R = 0,1) in the f
0*-* 1*-* 0*-* 1*-*
*----]A]--------]B]----* *----]A]--------]B]----*
] *-* *-* ] ] *-* *-* ]
*-* *-* *-* *-------> *-*
]F] 0,3 ]C] ]F] T 3,0 ]C]
*-* <-------* *-* *-* *-*
] *-* *-* ] ] *-* *-* ]
*----]E]--------]D]----* *----]E]--------]D]----*
*-* 3*-* *-* 3*-*
3) Station D reserves a higher priority 4) Station A removes its fra
of 3, overriding B's reservation (P,R=0,3). token at P,R = 3,0. A en
0*-* 1*-* 0*-* 1*-*
*----]A]--------]B]----* *----]A]--------]B]----*
] *-* *-* ] ] *-* *-* ]
*-* *-* *-* *-* *-*
]F] T 3,1] ]C] ]F] 3,1 ]C]
*-* <-* *-* *-* <-------* *-*
] *-* *-* ] ] *-* *-* ]
*----]E]--------]D]----* *----]E]--------]D]----*
*-* 3*-* *-* 3*-*
5) Station B repeats priority token 6) Station D transmits its p
making a new reservation (P,R = 3,1). frame with P,R = 3,1.
0*-* 1*-* 0*-* 1*-*
*----]A]--------]B]----* *----]A]--------]B]----*
] *-* *-* ] ] *-* *-* ]
*-* *-* *-* *-------> *-*
]F] T 3,1 ]C] ]F] T 1,0 ]C]
*-* <-------* *-* *-* *-*
] *-* *-* ] ] *-* *-* ]
*----]E]--------]D]----* *----]E]--------]D]----*
*-* 3*-* *-* 3*-*
7) Station D removes its frame upon return and 8) Station A originates a to
generates a token at the priority just used level 1 (as reserved) wit
(P,R = 3,1). Station A is still in Priority- and stays in Priority-Hol
Hold state awaiting a token of the priority the priority ultimately d
awaiting a token of the priority
-------------------------------------------------------------------------------
Figure 19. Token-Passing Ring Protocol - Access Priority, No Early Token-Releas
2.2.2.4.7 Additional Token-Passing Ring Considerations.: Using an average
frame size of 1000 bits to simulate the performance of a 4 Mbps token-passing
ring LAN with 100 active LAN devices results in a maximum throughput of about
3.6 Mbps. The token-passing ring protocol appears to be particularly stable
and even most efficient under high load conditions.
The impact of increased transmission speeds, increased numbers of attached
stations, or increased transmission distances on a token-passing LAN is
significantly less than similar changes on a CSMA/CD LAN. Because each
station regenerates the signal, increased distances are easier to support,
while transmission speed is primarily limited by the choice of media. The use
of bridges to provide additional device capacity and/or distance is an
attractive growth option because the absence of collisions simplifies the
processing requirements of bridges and maintains the deterministic
characteristics of the protocols.
In a token-passing ring, fairness in the access protocol and high priority
utilization by the bridge helps avoid frame loss. Even when a frame is
rejected due to bridge congestion, successful recovery is simplified by the
protocol.
2.2.3 Basic Token-Passing Bus Concepts
The TOKEN-PASSING BUS MEDIUM ACCESS PROTOCOL is differentiated from the
token-passing ring protocol mainly by topology requirements and by a
timer-based token management approach.
In a bus topology there is no physical sequence between stations. To be able
to pass the token from one station to its successor station in a logical ring
fashion, specific medium access function must be provided to initiate and
maintain the logical ring sequence allowing non-disruptive station insertion
or removal.
Instead of relying on a token monitor function as in the token-passing ring
access protocol, token management is distributed among all stations involved
in token-passing based on specific timer values and a continuous measurement
of the traffic load on the bus.
Some stations which do not require the ability to transmit frames are said to
be NON-TOKEN HOLDING stations. They can only receive frames and are bypassed
by the access protocol at token-passing time. These stations are only allowed
to respond to requests for acknowledgement.
As in any token-passing protocol, a station is only allowed to transmit data
while it holds the token. Possession of a token can therefore be described as
the right to transmit.
In a token-passing bus, the appearance of a logical sequence between stations
is provided by a concept of predecessor and successor stations. Each station
on the medium capable of participating in token holding establishes the
identity of its LOGICAL predecessor and LOGICAL successor station
(independent of physical arrangement on the medium) on the basis of the
descending numerical order of MAC addresses of each station. Thus the
logical sequence of the stations in Figure 20 under heading "2.2.3 Basic
Token-Passing Bus Concepts" would be A - C - D - B.
+------------------------------------------------------------------------------
]
]
]
] *-----------* *-----------*
] ] A ] P = B ] B ] P = D
] ] ] S = C ] ] S = A
] *-----------* *-----------*
] *-* ] ADDR = 99 ] ADDR = 11
] ] ]------------------------------------------------------------
] ] ]-----------------------------------------------------------]
] *-* ] ADDR = 66 ] ADDR = 33
] *-----------* *-----------*
] ] C ] P = A ] D ] P = C
] ] ] S = D ] ] S = B
] *-----------* *-----------*
]
]
] P = Predecessor Station
] S = Successor Station
+------------------------------------------------------------------------------
Figure 20. Token Bus Logical Sequence of Stations
The token is normally passed from station to its successor station using a
short TOKEN PASS FRAME. If a station fails to pick up the token, the sending
station uses a sequence of increasingly comprehensive recovery procedures to
find a successor station.
2.2.3.1.1 Frame Format: Figure 21 under heading "2.2.3.1.1 Frame Format"
shows the frame format as transmitted on a token-passing bus, including the
format of a token frame. Note the absence of access control and frame status
fields as present in a token-passing ring MAC frame.
+------------------------------------------------------------------------------
]
]
] * Generic MAC frame format Physical
] Trailer
] ]<------------- Physical Header -------------->] ]<-------->
] ] ]<-------- Code Violation Protection ----------->]
] ] ]<-------------- FCS Protection ---------------->]
] *---------------------------------------------------------------------
] ] Preamble ] SD ] FC ]Destination ]Source ] Data_Unit ] FCS ] ED
] ] ] ] ]Address (DA)]Address (SA)] ] ]
] *---------------------------------------------------------------------
] ] ] ] ]
] ] Frame Control Frame Check Sequence ]
] ] FF]CCCCCC (4 byte CRC value) ]
] ] ]
] Starting Delimiter Ending Delimite
] VV0VV000 VV1VV1]I]
]
] * Token frame format
]
] *--------------------------------------------*
] ] Preamble ] SD ]00001000] DA ]SA ] FCS ] ED ]
] *---------------------------------------------
]
] * Abort Delimiter *---------*
] Format ] SD ] ED ]
] *---------*
]
]
]
+------------------------------------------------------------------------------
Figure 21. IEEE 802.4 MAC Frame Format
o Preamble - a number of padding bytes used in broadband transmission by the
receiver to acquire signal level and phase lock using a known pattern. It
also guarantees a minimum End Delimiter (ED) to Start Delimiter (SD) time
period to allow processing of the frame previously received.
o SD, ED - see token-passing ring frame format
o FC - FF = B'00' - MAC control frame, CCCCCC = frame type
- B'000000' = Claim Token
- B'000001' = Solicit_Successor_1
- B'000010' = Solicit_Successor_2
- B'000011' = Who_Follows
- B'000100' = Resolve_Contention
- B'001000' = Token
- B'001100' = Set_Successor
o FC - FF = B'01' - LLC_data frame, B'10' = Station_Management frame, B'11'
= reserved.
CCCCCC = MMMPPP, PPP = LLC sublayer message priority and MMM = MAC action:
- B'000' = Request_With_No_Response
- B'001' = Request_With_Response
- B'010' = Response
o SA, DA - 48-bit MAC address fields, see Figure 13 under heading "2.2.2.1
MAC Addressing".
o Token frame format - a token is transmitted as a specific MAC control
frame with DA = unique MAC address of the successor station and a null
data_unit.
When a frame is received by a station, it is checked for transmission errors.
Invalid frames are not passed to the LLC sublayer. A MAC frame is invalid if
any of the following conditions occurs:
o The frame does not contain an integral number of bytes.
o The result of the cyclic redundancy check upon arrival does not match the
contents of the FCS field.
2.2.3.1.2 The Token-Passing Bus Protocol: The major token-passing bus MAC
functions cover:
o Ring initialization (Claim Token MAC frame)
o Station addition (Solicit_Successor frame)
o Station removal (Set_Successor frame)
o Recovery and management, including:
- Bus idle conditions
- Token-passing failures
- Duplicate token
- Detection of a station with a faulty receiver
- Duplicate addresses.
When the network is first powered up, or after a catastrophic error, the
logical ring needs to be initialized. This process is triggered by the
BUS_IDLE TIMER expiring in a LAN station. The detecting station sends a CLAIM
TOKEN MAC control frame.
As described above, each participating station knows the address of its
PREDECESSOR, the station that transmitted the token to it and its SUCCESSOR,
the station to which the token should be sent next.
After a station has completed transmitting data frames and has completed other
logical ring maintenance functions, the station passes the token to its
successor by sending it a token MAC control frame. Any failure in reaching a
successor station triggers a staged recovery mechanism, using other MAC
control frames (SET_SOLICITOR_1 and SET_SOLICITOR_2). If the token holder
does not receive a valid token after sending the token the first time, it
repeats the token pass operation. If the successor doesn't transmit after a
second token frame, the sender assumes that its successor has failed and sends
a WHO_FOLLOWS MAC control frame containing its successor's address in the data
field. The station detecting a match between its predecessor and the address
in the Who_Follows frame data field will respond by sending its address in a
SET SUCCESSOR MAC control frame. In this way, the token holding station
establishes a new successor excluding the failing station from the logical
ring.
Stations are added to the logical ring through a controlled contention process
using RESPONSE WINDOWS, a specific interval of time common to all stations,
and based on numerical address comparisons. The actual procedure is referred
to as the SOLICIT_SUCCESSOR PROCEDURE.
This procedure however raises a concern with respect to excessive delay
experienced by a station before gaining access to the LAN when many stations
attempt to insert into the logical ring and perform the solicit successor
procedure. The time a station has to wait between successive passes of the
token is called the TOKEN ROTATION TIME. To avoid an excessive TRT, every
station measures the rotation time of the token. If the time exceeds a
predefined value set by station management, the station will defer initiation
of the solicit successor procedure, and verify at the next appearance of the
token whether it is now rotating fast enough to perform the procedure.
A station can remove itself from the logical ring simply by not responding
anymore to the token passed to it. Ring station sequence recovery procedures
will adjust the successor and predecessor information in the predecessor and
successor stations respectively.
A more efficient way of leaving the logical ring is to have the exiting
station send a SET SUCCESSOR MAC control frame to its predecessor, containing
the address of its successor.
2.2.3.1.3 Access Priority: The token-passing bus protocol provides an
optional 8-level priority mechanism by which higher layer data frames, LLC
sublayer or higher level protocols, are assigned to eight different SERVICE
CLASSES according to their desired transmission priority. Service classes
range from 0 (low) to 7 (high).
At the access method level, there are four ACCESS CLASSES, corresponding to
four request queues to store pending priority frames. Access classes are 0
(lowest), 2, 4 and 6 (highest priority).
Token bus MAC maps the service class requested by the LLC sublayer into a MAC
access class (service classes 0 and 1 into access class 0, service classes 2
and 3 into access class 2, etc.).
The purpose of this priority mechanism is to allocate bandwidth to the higher
priority frames and to send lower priority frames only when there is
sufficient bandwidth left. To prevent any station from monopolizing the medium
with access class 6 frames, a station can only send class 6 frames for a
maximum time defined by station management. When this time expires a station
must release the token even if it has additional class 6 frames ready to
transmit.
Similarly, each access class is assigned a TARGET TOKEN ROTATION TIME. For
each access class a station measures the time it takes the token to circulate
the logical ring. If the token returns to a station in less than the target
token rotation time, the station can transmit more frames of that access class
until expiration of the TTRT. If the token takes more than the TTRT for a
specific access class, no frames of this priority class can be sent at this
pass of the token. The actual algorithm consists of loading the residual value
of a token rotation timer into a TOKEN_HOLDING TIMER. If an access class's
queue is empty or if the token_holding timer expires, the station begins to
serve the next lower access class. When the lowest priority class is serviced
the station performs any required logical ring maintenance operation and
releases the token for its successor station.
2.3 Fiber Distributed Data Interface Concepts
Growing use of personal workstations has meant the increased use of file and
print servers and communication gateways, to share resources between users and
hence reduce costs. Applications have been developed that turn these concepts
into real benefits, enabling data sharing between workstations, departmental
processors and large mainframe hosts. As more users clamor to take advantage,
the bandwidth requirement on the LAN keeps growing, if service levels or
response times are to be maintained.
Yet new application areas are now being developed. More sophisticated
workstations with high-resolution graphic capabilities are very suitable for
image display, enabling documents to contain pictorial information as well as
simple text. The amount of data that is needed to represent a single page of
such a document, even with efficient data compression techniques, can approach
the megabit size; moving it around takes a high bandwidth network. As
technology continues to improve, and becomes available at cost effective
prices, yet more sophisticated applications will be developed.
If new applications are to be placed alongside the applications of today, then
another order of magnitude increase in available bandwidth will be required
within the establishment. The LAN that will provide this bandwidth is FIBER
DISTRIBUTED DATA INTERFACE or FDDI.
2.3.1.1.1 IBM and FDDI: On June 19, 1990, IBM announced a Statement of
Direction, that said:
o IBM would offer high-performance FDDI LAN, workstation, host attachment
products and interconnection between FDDI, token-ring and Ethernet LANs
o That these products would conform to the emerging ISO 9314 standards.
o The most commonly used fibers, including the IBM Cabling System, would be
supported
o That network management solutions for the FDDI environment would be
provided
o Single-mode fiber and LASER technologies for longer transmission distances
would be offered.
When IBM announced the RS/6000 processor in early 1990, a statement was made
that that machine would eventually support connection to an FDDI LAN. The IBM
3172 Interconnect Controller Model 002 is the second product to contain an
FDDI attachment content.
Since there are no IBM products that can currently attach to an FDDI network,
it is not possible to describe any IBM implementation. For the benefit of
those readers who wish to understand a little more about FDDI, what it looks
like, and how it works, the next sections describe the architecture as
currently defined within the standards.
2.3.1.1.2 History: FDDI has evolved from the activities of the AMERICAN
NATIONAL STANDARDS INSTITUTE (ANSI) committee X3T9.5, which began its quest
for a high-speed, optical fiber based communications network in 1982.
Originally envisioned as a standard for future host-channel requirements, it
rapidly became viewed as the next generation LAN standard, improving on, and
drawing techniques from, the IEEE work on lower-speed LANs (50Mbps).
As of today, the full protocol has not yet become a standard, and delay in
this area has meant that a LAN implementing a full standardized FDDI cannot be
built. The following paragraphs will give an overview of the structure,
characteristics and protocols of this architecture, and attempt to position
today's token-ring technology with regard to this important new architecture.
2.3.1.1.3 Structure: FDDI describes a dual counter-rotating fiber-optic ring
operating at 100Mbps, in many ways similar, and in some ways different, to the
IEEE token-ring. FDDI uses a token-passing protocol, with each station having
the chance to transmit frames when a token passes. The algorithm that a
station uses to decide how many frames can be transmitted is described in
"2.3.1.1.6 Allocating FDDI Capacity." A difference between token-ring and FDDI
is that an FDDI station can transmit multiple frames one after the other
without releasing a token in between. In fact, the token-ring standard allows
token-ring stations to do this also. It is an implementation decision as to
whether to do this or not. In order to allow fairness of medium access to
token-ring stations, IBM has chosen not to allow multiple frames without token
release. FDDI has other methods to ensure that stations get access to the
ring, so multiple frames are allowed.
In token-ring, all stations with the exception of the IBM 8220 Optical Fiber
Converter and the IBM 8230 Controlled Access Unit attach to the ring via an
access unit, which may be active or passive. In FDDI, stations can attach
directly on to the ring or through a concentrator. A concentrator is a device
similar in concept to an active token-ring access unit, in that it provides
for attachment to the rings for many devices connected to lobes. However, with
FDDI, the stations use the fiber-optic media to attach to the concentrator,
not copper cabling as with token-ring. Since fiber-optics cannot carry a
PHANTOM CURRENT as is used to switch the relays within the token-ring access
units, insertion of the device into the ring is controlled by a defined
protocol.
FDDI uses two rings, one called the PRIMARY and the other the SECONDARY ring.
Compared to the token-ring, the primary ring may be likened to the main-ring
path, with the secondary ring acting as the backup ring path.
Stations, if attaching directly, or concentrators can attach to either or both
of these rings. In order to differentiate between stations that attach to one
or both rings, two classes of station are defined.
A CLASS A station attaches to both of the rings directly, while CLASS B
stations attach to only one of the rings. Attachment of class B stations can
be direct or through a concentrator.
A class A station might be a host gateway or perhaps a bridge to a lower speed
LAN. For these stations, availability rather than cost would be the prime
design consideration, however there is no requirement that these types of
station must be class A. One important fact about a class A station is that
it should not be powered off. If it does, then, as will be seen later, the
ring will reconfigure around it in order to remain operational. If two class A
stations were absent from the ring, then the ring would split into two. This
is a similar situation to that which can happen in the token-ring if two sets
of IBM 8220 Optical Fiber Converters were both broken at the same time. Host
gateways or LAN bridges might be expected to be continuously available, an
attribute of a class A station. Ordinary workstations, such as personal
computers, which will be powered on or off would be suitable for class B
attachment. The cost of attachment of class B stations would also be less,
again making the approach suitable for the mass of customer installed
workstations.
Concentrators, in addition to the stations, are able to attach to either or
both rings.
A DUAL ACCESS CONCENTRATOR (DAC), as its name implies, attaches to both rings,
while a SINGLE ACCESS CONCENTRATOR (SAC) attaches to only one. Only Class B
stations can attach to a concentrator. A diagram illustrating this is shown in
Figure 22 under heading "2.3.1.1.3 Structure". The figure shows class A
stations attached directly to both rings. A dual access concentrator (DAC) is
shown, supporting some class B station attachments, as well as a single access
concentrator (SAC) treed off of it. Also attached to the primary ring path,
are a direct attached class B station, and a tree of two other SACs.
Direction of token rotation is shown by the arrows.
-------------------------------------------------------------------------------
*-------* *-------* *------------*
] Class ] ] Class ]---] token-ring ]
] A ] Primary ring ] A ] *------------*
*-----+-------+---------<--------------+-------+----------*
] *---+-------+--------->--------------+-------+--------* ]
] ] *-------* Secondary ring *-------* ] ]
] ] Gateway Bridge ] ]
V A V A
] ] ] ]
] ] ] ]
] ] *---------------* ] ]
] *---+---------------+-------------<-------------------* ]
] ] DAC ] *-------------* ]
*->---+-**------**----+--->----**--->--+--**->--**---+----*
*-++------++----* ]] *--++----++---*SAC
]] ]] ]] ]] ]]
]] *-++-----------* ]] ]] *-++-------------*
]] ] ]*---<-----* ] ]] ]] ] ]*-----<-----* ]
]] ] V SAC ] ] ]] ]] ] ] SAC ] ]
VA ] *-**->--**-* ] VA VA ] *-**--->--**-* ]
]] *---++----++---* ]] ]] *---++------++---*
]] ]] ]] ]] ]] ]] ]]
Bridge ]] ]] ]] ]] ]] ]] ]] Bridge
*-------* *-------* VA *-------* *-------* VA *-------*
]Class B] ]Class B] ]] ]Class B] ]Class B] ]] ]Class B]
*-------* *-------* ]] *-------* *-------* ]] *-------*
] ]]Bridge +- ]] ]
----+--------+---- ]] ]Ethernet *-------* *----------*
Ethernet *-------* ] ]Class B] ]token-ring]
]Class B] -+- *-------* *----------*
*-------* ]
-------------------------------------------------------------------------------
Figure 22. FDDI - Station attachment
A class A station or a dual access concentrator has a medium access control
instance on the primary only and optionally on both the primary and secondary
ring path. The significance of this will become clearer during the discussion
on ring recovery.
During normal ring operation, the secondary ring is idle with no user traffic
being sent around it, as is the backup path in token-ring implementations.
In order to utilize an FDDI ring as an establishment or campus backbone for
interconnecting other LAN types, perhaps Ethernets and token-rings, bridges
will be needed to link the slower-speed LANs to the backbone. The structure of
an FDDI frame is different from both token-ring and Ethernet frames. The FDDI
frame structure is illustrated in Figure 24 under heading "2.3.1.1.5 Medium
Access Control Layer (MAC)", but for the moment, the important point is that
the frame check sequence characters for the frame as it exists on the
token-ring or Ethernet will be different from that on the FDDI LAN. This
means that the bridges will have to recalculate it as the frame passes from
one LAN to the other. Although not necessarily important from a performance
point-of-view, it does mean that there could be a data integrity exposure
within the bridge. There would be a short time when the frame was not
protected by a frame check sequence. A bit dropped or picked up would lead to
the frame being carried end-to-end without the possibility of the error being
detected.
Another problem in obtaining complete inter-operability between Ethernet and
token-ring attached stations is that of source and transparent bridging
techniques being used by the different LANs. This could be solved by the
concept of the SRT bridge. This new design for a bridge is discussed in "6.3.3
Source Routing/Transparent Bridging Inter-operability."
2.3.1.1.4 Architecture: One of the prime requirements for FDDI is as a
HIGH-SPEED ESTABLISHMENT BACKBONE to interconnect heterogeneous LANs. To this
end, the FDDI LAN design must have reliability and capacity built in.
Attachment cost does not have to be so carefully limited, since most of the
cheaper workstations, would be connected to the lower speed LANs that were
themselves bridged to the FDDI backbone.
FDDI has satisfied these requirements by:
o Operating at a data rate of 100Mbps, providing lots of capacity
o Having a dual optical fiber ring configuration, that can be reconfigured
by the class A stations in the event of a failure. This provides for
reliability and recovery.
o Allowing stations to attach to both of the rings (more expensive) or only
one (cheaper).
ANSI, in defining and developing the standard was able to make much use of the
work already done by the IEEE. To this extent, they used a similar model for
the layering of the protocols. The structure is shown in Figure 23 under
heading "2.3.1.1.4 Architecture".
-------------------------------------------------------------------------------
A *---------------------------*
] ] IEEE 802.2 ]
OSI ] ] Logical Link Control ]
Layer 2 ] ]---------------------------+--------------*
] ] MAC ] ]
V ] Medium Access Control ] ]
- - - ]---------------------------] ]
A ] PHY ] Station ]
OSI ] ] Physical Protocol ] Management ]
Layer 1 ] ]---------------------------] ]
] ] PMD ] ]
] ] Physical Medium Dependent ] ]
V *------------------------------------------*
-------------------------------------------------------------------------------
Figure 23. FDDI Architecture
The FDDI standard assumes the use of the IEEE 802.2 logical link standard; it
does not attempt to define it as part of FDDI.
2.3.1.1.5 Medium Access Control Layer (MAC): In this layer, FDDI borrows
very extensively from the techniques used by the IEEE, particularly the 802.5
token-ring specification.
FDDI uses a token access protocol; stations are aware of their nearest active
upstream neighbor, and with Class A stations, their neighbor on the secondary
ring as well. There are similar processes for ring monitoring, such as the
CLAIM TOKEN and BEACONING PROCESS.
As with token-ring, when an FDDI station wishes to transmit a frame, it must
wait for a token to arrive. In the token-ring architecture, the transmitting
station flips a bit in the token to indicate that it is a token no longer, and
appends the data to this frame header. With FDDI, the token is removed from
the ring by the transmitting station before actual frame transmission begins.
The frame size on an FDDI LAN is limited by clocking constraints to 4500
bytes. At the FDDI data rate, and given the fact that an FDDI ring can be
physically very large, it is extremely likely that the frame will be
physically shorter than the length of the ring. For token-ring at 16Mbps, IEEE
802.5 has specified a concept known as EARLY TOKEN RELEASE which is designed
to prevent wasting bandwidth on the ring. This has been described in "2.2.2.2
Early Token Release" . FDDI uses a similar concept. A token is transmitted
immediately behind the last data frame, whether or not the head of the frame
has returned to the sending station. This token is available for any other
station to use.
The frame and token structure on an FDDI LAN is shown in Figure 24 under
heading "2.3.1.1.5 Medium Access Control Layer (MAC)".
-------------------------------------------------------------------------------
FDDI FRAME
SFS FCS Protected EFS
<----------->]<-------------------------------->]<------->
*----------------------------------------------------------*
] PA ] SD ] FC ] DA ] SA ] Information ]FCS] ED ] FS ]
*----------------------------------------------------------*
PA Preamble - used for clock synchronization - 64 bits
SD Starting delimiter -
FC Frame Control
DA Destination Address
SA Source Address
FCS Frame Check Sequence - 32 bit result of a CRC check
ED Ending Delimiter
FS Frame Status - address recognized, frame copied and error detected
bits
SFS Starting Frame Sequence
EFS Ending Frame Sequence
FDDI TOKEN
*-----------------------* FC = 1000 0000 Unrestricted token
] PA ] SD ] FC ] ED ] FC = 1100 0000 Restricted token
*-----------------------*
FRAME CONTROL
*-------------*
] CLFF ] ZZZZ ]
*-------------*
C - Class bit 0 = Asynchronous 1 = Synchronous
L - Length of address fields 0 = 48 bit 1 = 16 bit
FF - Format 00 = MAC or SMT frame
01 = LLC frame
10 = Implementation dependent
11 = Reserved
ZZZZ - Control bits Dependent on frame type
-------------------------------------------------------------------------------
Figure 24. FDDI Frame and Token Formats
Stations on an FDDI ring can use 16 or 48 bit addressing. If a station uses 16
bit addressing, it must be able to recognize the 48 bit all-stations broadcast
address of a 48 bit station. The converse is also true for a 48 bit station.
Stations addressed in the two different ways must also be able to participate
in all the station management (SMT) and MAC protocols on the ring.
2.3.1.1.6 Allocating FDDI Capacity: Because the token is always released
before the frame header has returned, the priority reservation scheme that is
used in the token-ring architecture to give stations priority access to the
ring cannot work in FDDI. Additionally, FDDI is designed to give far greater
control over the use of the ring capacity to connecting stations. The
mechanism that is used is heavily dependent on timers within each station.
FDDI defines two classes of traffic:
SYNCHRONOUS
for which a guaranteed bandwidth is required; that is a station must
gain access to the ring, and transmit its frames within a specified
time period. Such an application might be a manufacturing control
application, where a robot arm needs to be accurately controlled by
a processor.
ASYNCHRONOUS
traffic will only be transmitted when the load on the ring drops
below a specified level. Such traffic might be file transfer where
overall access or response time is not quite such an issue.
To control the amount of each kind of traffic that can be transmitted by a
station, FDDI implements a TIMED TOKEN ACCESS PROTOCOL. Stations expect to see
a token within a specified time interval. This interval is known as the TARGET
TOKEN ROTATION TIME (TTRT). All stations have the same value for TTRT set
within them. The value is determined when a station initializes itself on the
ring.
Each station measures the time between successive appearances of a token. The
stations uses a TOKEN ROTATION TIMER (TRT). to do this.
When a token passes a station, the TRT is set to the value of the TTRT and
starts to count down. If the TRT timer expires before the next token is seen,
a counter called the LATE COUNTER is incremented. The token is said to be
late. Under normal circumstances, the token should reappear within the target
time. The token is then said to be EARLY.
Each time a station sees a token, three things happen:
o The station may start to transmit synchronous frames
o If the TRT has not expired, that is the token was early, the remaining
time is stored in the TOKEN HOLDING TIMER (THT). The THT is therefore the
amount of time by which the token was early.
o The TRT is reset to the value of TTRT and allowed to start running down
again.
The station is allowed to transmit synchronous frames until a time allocated
for synchronous data transmission expires (the synchronous allocation timer).
This timer value may be different at each station (it may be zero at some
stations if they have not been enabled by an application for synchronous
transmission), but the sum value of all the synchronous allocation timers on
active stations must always be slightly less than the target token rotation
time. It is up to the station management protocols to make sure that this is
so.
When the timer expires, or all synchronous frames are transmitted, the station
may be allowed to transmit asynchronous frames. The decision is based on the
value of the LATE COUNT. If the count is at zero, asynchronous frames can be
transmitted for the length of time that was stored in the token holding timer.
When this timer runs down to zero, transmission must stop, and a token must be
released.
Meanwhile, the TRT is steadily running down. It has been doing so during both
synchronous and asynchronous transmission. If both of these transmissions were
stopped because of timer expiration (synchronous allocation and token holding)
and assuming that other stations have frames to send, the TRT will almost
certainly expire before the token is seen again. The token is late; the late
counter will be incremented, there will be no value to put in the token
holding timer, and the station will not be allowed to send any asynchronous
frames the next time it does see the token. In a way, the station has paid a
penalty for having had so much traffic to send. This penalty only applies to
asynchronous traffic however; a station may always transmit synchronous
traffic when it sees a token.
When the token next arrives early, the late counter is decremented. When the
value gets to zero, the station will be able to transmit asynchronous traffic
once more.
When this operation is considered across all stations on the ring, the
capacity allocation algorithm is attempting to keep the actual token rotation
time less than the target. Since synchronous traffic can always be sent, the
mechanism guarantees an amount of bandwidth to the synchronous traffic.
Asynchronous traffic is only sent when there is spare capacity on the LAN, and
fairness of station access is maintained.
In order to summarize this algorithm, we will present an example. Suppose
that:
o TTRT is set at 80 milliseconds
o The synchronous allocation timer is set to 10 milliseconds
Now we will follow it through:
1. A token arrives
2. TRT is set to TTRT (TRT contains 80)
3. No traffic to send, so token is released
4. Traffic starts to get queued for sending within the station
5. The token reappears - TRT has got down to 30, that is there was an
interval of 50 milliseconds between step 1 and 4.
6. The token is early
7. The value of 30 is placed in the token holding timer
8. TRT is reset to 80 and starts to run down
9. Synchronous traffic is sent for 10 milliseconds (synchronous allocation
timer)
10. TRT now reads 70
11. Asynchronous traffic is sent for 30 milliseconds (token holding timer)
12. TRT now reads 40
13. A token is released
14. Last time, the token took 50 milliseconds to go around the ring. If any
other stations have traffic to transmit, it will take longer.
15. TRT expires, the late counter is incremented to a value of 1.
16. More asynchronous traffic is queued within the station
17. TRT is reinitialized to 80, and runs down
18. The token reappears 15 milliseconds later
19. This time the token has taken 55 (40+15) milliseconds to go around. TRT
has the value 65 in it. The token appears to be early, but the late count
still has a value of 1 in it, so the token is late. The station can
transmit synchronous traffic, but no asynchronous. If there is no
synchronous traffic to send, the token passes by.
20. The late count is decremented to zero
21. The token reappears after 50 milliseconds
22. TRT now has a value of 15 in it, the late count is zero, so the station
can transmit its asynchronous traffic for 15 milliseconds.
For asynchronous traffic two other mechanisms may also operate. These are:
o Restricted token
This allows two stations, communicating with asynchronous traffic to take
all the asynchronous bandwidth available on the LAN. When a station
wishes to initiate this, it transmits its asynchronous frames, and
releases a RESTRICTED TOKEN. Only the last station to receive an
asynchronous frame can use the restricted token for asynchronous traffic,
so the two stations can go on transmitting to each other. The restricted
token is useable by any station with synchronous traffic to send.
Therefore the guarantee of synchronous bandwidth remains.
o Priority
Asynchronous traffic can be allocated to one of eight priority levels.
When there is time available for asynchronous frame transmission, the
frames are sent in priority order, level 7 down to 0.
Using all these mechanisms, FDDI is able to cope with traffic needing a
defined amount of bandwidth, and also support the "bursty" traffic whose
response time requirement is less defined. It also allows for those
applications where a prolonged and restricted communication between partners
is required.
2.3.1.1.7 Physical Layer (PHY): As in the IEEE standards, the physical layer
specifies the signalling rate, encoding and clocking of the LAN. For FDDI,
the data rate on the ring is 100Mbps.
The IEEE token-ring standard specifies the use of Differential Manchester
Encoding. This method of coding means that the signal frequency on the ring is
twice the nominal data rate. For a 16Mbps token-ring, this rate would be 32
million signalling elements a second or 32M baud. If this coding technique
were used for 100Mbps operation, this would mean that the signalling rate
would have to be 200Mbaud, which would put additional cost implications on the
technology used and drive distances obtainable.
So, a different technique is used. The datastream is transmitted as a series
of 4 bit symbols, with each symbol being encoded within a 5 bit pattern. This
is known as 4B/5B encoding, and means that the signalling rate need only be
125M baud. Since 4 bits represents 16 digits, and 5 bits 32 digits, 16 of the
5 bit codes are redundant. These codes can be used for special purposes, such
as delimiting the start and end of frames.
In order to improve reception of the symbols by a station, these 5 bit
patterns are additionally encoded using NON-RETURN TO ZERO INVERTED (NRZI).
This technique means that a bit ( 0 or 1 ) is indicated by not changing (0) or
changing (1) the polarity of the signal (light on or off) at each bit clock
time. This differential technique improves the accuracy of signal reception,
as well as helping to maintain synchronization between transmitter and
receiver.
In the token-ring, master clocking is done by the ACTIVE MONITOR. In the FDDI
standard, responsibility for the clocking rests with all stations, primarily
because the clocks need to run very fast. The problems of synchronizing
individual clocks to a master would lead to expensive circuitry. Therefore
each station has an autonomous clock, independent of all others, with
synchronization maintained by the NRZI coding technique.
FDDI defines a maximum of 500 stations attaching per ring.
2.3.1.1.8 Physical Medium Dependent Layer (PMD): This layer is new if
compared with the IEEE standards. It is used to specify the fiber type and
size used by FDDI, the connectors, and the wavelength of the light beam that
is transmitted down the fiber. The standard fiber is a multi-mode fiber of
62.5/125 micron, (10(-6) meters) although fiber sizes of 50/125, 82.5/125, and
100/140 microns are listed as alternatives, as well as single-mode fiber.
By selecting a wavelength of 1300nm (nanometers or 10(-9) meters), the Light
Emitting Diode (LED) is still suitable as an optical transmission device. The
wavelength is not the same as that used by the IBM 8219 or 8220 optical fiber
converters (850nm); at the FDDI signalling rate that wavelength would begin to
lose transmission distance. Use of a longer wavelength (1550nm) would require
a more expensive laser transmitter, so the selection of 1300nm does mean that
the cheaper LED technology can still be used.
FDDI has defined that the maximum fiber drive distances between stations shall
be 2 kilometers (2000 meters). If a ring contained all 500 stations at the
maximum distance apart, the ring would span 1000 kilometers.
2.3.1.1.9 Station Management (SMT): Station Management is perhaps the single
most important function within the FDDI architecture and one to which IBM has
made significant contributions within the standards-making body. It defines
the protocols that stations use to intercommunicate, so as to set their
timers; the target token rotation timer or their synchronous allocation timers
for example. Without a consistent set of protocols which are used by all
stations that attach to the ring(s), it is impossible for FDDI to deliver the
function, capacity and performance for which it was designed.
Station management is also responsible for describing what the MANAGED OBJECTS
within an FDDI station should be. A discussion on these concepts can be found
in "2.4 LAN Station Management."
2.3.1.1.10 Ring Reconfiguration: As with the token-ring, the secondary FDDI
ring is available to establish an alternate ring path should a break or other
disruption occur on the network. For example, a failing or inactive
dual-attached station or concentrator will be detected by adjacent units both
upstream and downstream of the failed station. The stations detecting the
failure can wrap the primary path to the secondary path, thus completing the
ring via the secondary path, and bypassing the failed station or concentrator.
This is shown in Figure 25 under heading "2.3.1.1.10 Ring Reconfiguration".
The default mode of operation is for a dual-attached station to execute a wrap
upon detection of an error on either the primary or secondary path. Although
the secondary ring is idle in normal operation, the optional secondary ring
MAC instances contained within the dual attached stations are able to detect
failures. The situation is dissimilar to that seen in the token-ring
environment with the IBM 8220 Optical Fiber Subsystem, and the IBM 8230
Controlled Access Unit. The Station Management function can detect certain
types of physical errors, such as broken fiber, faulty connection and a bad
receiver or transmitter, without a secondary ring MAC instance.
-------------------------------------------------------------------------------
*------------------------*
] Class A Node ]
] Wrap ------* ]
] *-* --* ] ]
] ] ] ] ] ]
*------------------------*
*-----* *-----*
] ] ] ]
Concentrator ] ] ] ]
*---------* *-------* ] ] ] ]
] Class B ] ] ] PI V A ] ]
] Node ]-* *-] ]-* ] ] ] ]
] *-] ]--<--] ]-------] ]--* ] ] ]
] *-] ]-->--] ]-* *-] ]----*
] ]-* *-] ] ] ]-* Break
] ] ] ] ] ] SO
*---------* ] ] ] ] ] ]
] ] ] ] ] ]
*---------* ] ] ] ] ] ]
] Class B ] ] ] ] ] ] ]
] Node ]-* *-] ] ] ] ] ] *---------*
] *-] ]--<--] ]-* ] ] ] ] ] Class A ]
] *-] ]-->--] ]-* ] ] ] ] *-] Node ]
] ]-* *-] ] ] ] ] *---] ]----* ]
] ] ] ] ] ] *-----] ]--* ] ]
*---------* ] ] ] ] *-] ] ] ]
] ] ] ] ] ] ] ]
*---------* ] ] ] ] Secondary Ring ] ]
] Class B ] ] ] ] ] ] ] ]
] Node ]-* *-] ] ] ]-* V *-] Wrap ]
] *-] ]--<--] ]-* *-] ]----------<----------] ]--* ]
] *-] ]-->--] ]-------] ]---------->----------] ]--* ]
] ]-* *-] ]-* A *-] ]
] ] ] ] ] ] ]
*---------* *-------* Primary Ring *---------*
-------------------------------------------------------------------------------
Figure 25. FDDI Dual Ring Reconfiguration
2.3.1.2 FDDI Standards
As has been stated, the FDDI architecture is divided into four functional
parts:
o Physical Medium Dependent (PMD)
o Physical (PHY)
o Medium Access Control (MAC)
o Station Management (SMT)
At the time of publication of this document, then:
o PMD is an International Standard - ISO 9314/3
o PHY is an International Standard - ISO 9314/1
o MAC is a Draft International Standard - DIS 9314/2
o SMT is being balloted as a draft ANSI standard.
2.3.1.2.1 Token-Ring Comparison: It should be apparent from the previous
text that there are many similarities between the FDDI standard and the IEEE
802.5 token-ring standard. There are also some differences, many of which are
required by the increase in speed of FDDI and hence timing considerations.
The differences are summarized in Figure 26 under heading "2.3.1.2.1
Token-Ring Comparison" .
-------------------------------------------------------------------------------
*-------------------------------------------------------------------*
] ] Token-Ring ] FDDI ]
]---------------------+----------------------+----------------------]
] Speed ] 4/16 Mbps ] 100 Mbps ]
] ] ] ]
] Standard ] IEEE 802.5 ] ANSI X3T9.5 ]
] ] ] ]
] Medium ] shielded copper ] optical fiber ]
] ] optical fiber ] ]
] ] ] ]
] Optical fiber ] 50/125 ] typical 62.5/125 ]
] ] 62.5/125 ] optional 50/125, ]
] ] 100/140 ] 50/100 and 9/125 ]
] ] ] ]
] Number of active ] max. 260 ] max. 500 ]
] stations ] ] ]
] ] ] ]
] Distance between ] max. 2km ] max. 2km (multimode)]
] stations ] (with converters) ] longer w/singlemode ]
] ] ] ]
] Max. frame length ] 17997 byte ] 4500 byte ]
] ] ] ]
] Data Encoding ] Manchester ] 4B/5B NRZI ]
] ] Differential ] ]
] ] ] ]
] Clocking Control ] Centralized ] Distributed ]
] ] ] ]
] Recovery ] More centralized ] Distributed ]
] ] ] ]
] Bandwidth allocation] Priority/reservation] Timed token ]
] traffic ] Asynchronous ] Synchronous ]
] ] ] Asynchronous ]
] ] ] ]
] Consecutive frames ] No - one per station] Multiple ]
] ] per token ] ]
] ] (IBM Implementation) ] ]
*-------------------------------------------------------------------*
-------------------------------------------------------------------------------
Figure 26. FDDI and Token-Ring - Architectural Differences
2.3.1.2.2 Migration: The FDDI standard is just beginning to stabilize enough
to allow a full implementation to be considered. Although some companies are
offering FDDI implementations at this time, it may be yet some time before a
common implementation can be adopted by all vendors. Many users may not have
an immediate need for the high bandwidth offered by FDDI, and may be wise to
consider a migration strategy that begins with the 16Mbps token-ring. Due to
the underlying similarity of the infrastructures (fiber type and topology),
migration would be simple once the FDDI standard is complete and stable.
A key factor to consider in migration to FDDI is network management. The IBM
LAN Network Manager, coupled with NetView, is one of the most advanced LAN
management tools available on the market today.
2.4 LAN Station Management
2.4.1 Open Systems Management
As has been described in "2.0 LAN Technology" LANs have varying degrees of
management capability built into them by virtue of their architecture.
Dependent particularly on their PHYSICAL TOPOLOGY and MEDIUM ACCESS CONTROL
(MAC) layer implementation they can offer various amounts of essentially
problem determination information as input to the management process. For
instance, LANs that implement the IEEE 802.5 token-ring standard are able to
indicate where a ring is broken because of the FAULT DOMAIN, a concept that
would be impossible without the inclusion of "ring poll" and "nearest active
upstream neighbor" (See "2.2.2.4 Neighbor Notification") in the MAC
architecture.
However, as the sophistication and power of stations attaching to LANs grows,
as well as the number of stations and hence applications running on them,
there is a growing need to provide much more management function than the
answer to the question: "Is the LAN broken, and if so, where?"
Additionally, LANs have many of the characteristics of open systems.
Standards bodies have been increasingly active during the architectural and
development phases of LAN technology to make sure that a particular
implementation is sufficiently well described so that it can be attached to
different manufacturers' equipment.
Within the scope of Open Systems Interconnection (OSI), network management has
been defined to include the following five areas:
o Fault Management
How much problem determination is needed to be done? It may be very useful
to know where the LAN is broken, or if it is suffering a high rate of
recoverable errors, but what happens if an application simply stops
communicating with its partner? How does the person responsible for
resolving problems proceed when the medium appears to be working
correctly?
Fault management may be summarized as detecting, analyzing, correcting and
tracking events and incidents in the network operation.
o Configuration Management
How does the management process know what the stations really are;
intelligent workstations, gateways, servers? How does it know what the
serial numbers of those machines are, where they are physically located,
and where they are plugged into the LAN? How does the management process
know what the current operational parameters of the network are?
How do applications know on which station their partner application
resides? Does this information have to be defined to them, or do they have
some way to dynamically find out, for example by using names? NAME
MANAGEMENT is an integral part of OSI configuration management.
This can be summarized as managing the physical and logical
"relationships" among the networking resources.
o Performance Management
How do we determine the utilization of the LAN by individual stations?
Which stations are the heaviest users of the LAN, and what applications
are they using? How do we make the decision as to whether to split the
application across multiple stations, or buy a more powerful machine?
Performance management collects performance data, tunes, distributes
workloads and enables capacity planning.
o Accounting Management
What is the basis for charging for the use of the common transmission
medium so that the costs of providing the service can be offset against
usage?
o Security Management
How is access to the LAN controlled? What is to stop someone from plugging
a station into the LAN and "stealing" the data flowing on that LAN? What
are the security implications?
In order to facilitate these functions, there is a need for some kind of
management application, working in the stations, to maintain information,
update counters, set and reset parameters, and to make this information
available to a central management machine. It would also be desirable if the
information kept at each station were itself standardized, so that different
station management applications (perhaps obtained from different vendors) had
identical function, and were thus able to provide coherent LAN-wide
information. As spontaneous events occur in these stations, then some form of
notification process is also required, such that a central LAN management
station can be kept aware of what is going on.
So, if the management function is to be readily available on any LAN station,
whatever its type and make, and an ability for a central management machine to
monitor all stations is to be provided, then it would seem sensible to adopt
techniques that are not vendor proprietary, but which follow those described
by OSI.
OSI Network Management is not completely defined for "system resources", (that
is the stations or applications); it is defined for the interconnection of
them, and provides a model for the management process.
It also provides a set of terminology that will be useful to understand before
continuing. The full set of definitions is well described in another ITSC
publication, ISO STANDARDS AND OSI/COMMUNICATIONS SUBSYSTEM IMPLEMENTATION
FOR OSI SYSTEMS MANAGEMENT AND THE DIRECTORY SERVICE, an "IBM Internal Use
Only" publication, but for the benefit of the reader, the relevant definitions
are given here.
The MANAGEMENT PROCESS of a communications environment is an
information-processing application. Since a network environment is
DISTRIBUTED, the individual components of the process are themselves
distributed. Management processes are categorized as:
o MANAGING PROCESS is that part of a distributed application process that
has responsibility for one or more management activities.
o AGENT PROCESS is that part of the distributed application process that,
at the request of a managing process, manages associated "managed
objects".
MANAGED OBJECTS are data communications resources that exist independently of
their need to be managed. Examples could include LAN attachment cards, logical
link connections, and timers.
MANAGEMENT OPERATIONS are effected through the use of "managed objects". The
agent processes perform the management functions upon receipt of directives
specifying management operations on managed objects.
NOTIFICATIONS are produced by the agents as the result of unsolicited events
that occur within the managed objects that they control.
This structure is illustrated by Figure 27 under heading "2.4.1 Open Systems
Management".
Open System 1
*-------------------------------*
] ]
] *--------------------* ]
] ] ] ]
] ] Managing Process ] ]
] ] ] ]
] *--------------------* ]
] A A ]
*-------------+--+--------------*
] ]
Management Operations ] ]
] ]
] ] Notifications
Open System 2 ] ]
*-------------+--+--------------*
] V ] ]
] *--------------------* ]
] ] ] ]
] ] Agent Process ] ]
] ] ] ]
] *--------------------* ]
] A ]
] ] ]
] V ]
] *-------------------------* ]
] ]Managed Objects ] ]
] ] *--* *--* *--* *--* ] ]
] ] ] ] ] ] ] ] ] ] ] ]
] ] *--* *--* *--* *--* ] ]
] *-------------------------* ]
] ]
] ]
*-------------------------------*
Figure 27. Interactions between Management Processes
OSI provides a protocol that allows an application running on a central
management machine to communicate with another application on a managed
machine in order to set parameters, obtain information and be notified of
spontaneous events. This protocol is called CMIP or COMMON MANAGEMENT
INFORMATION PROTOCOL and will be discussed below.
2.4.1.1.1 Common Management Information Protocol: CMIP is an Open Systems,
application layer protocol that provides the ability to send commands and
obtain responses between management applications. It provides a set of
services that themselves are defined by COMMON MANAGEMENT INFORMATION SERVICES
or CMIS.
These services are:
o GET - Get specific information and respond with it.
o SET - Set a parameter (and confirm that it has been done).
o ACTION - Perform some action(s) (and respond with the result).
o CREATE - Used to create thresholds (for instance error/traffic ratio).
o DELETE - The opposite of CREATE.
o EVENT - A notification that something has happened, that is an unsolicited
event, and hence is being reported. In certain cases, this notification
may require a response, and this is termed a CONFIRMED EVENT.
The CMIP protocol is itself conveyed by another OSI protocol, REMOTE
OPERATIONS (ROS). ROS provides a vehicle for the CMIP communication and itself
provides four commands. These are:
o INVOKE - Requests that an operation be performed.
o RESULT - Reports the successful completion of that operation.
o ERROR - The request was not completed successfully.
o REJECT - Used to reject the request.
CMIP/ROS uses connection-oriented services.
2.4.1.1.2 A Model for LAN Station Management Figure 28 under heading
"2.4.1.1.2 A Model for LAN Station Management" is an architectural model for
the internal management of LAN stations, that is the agent process. It is a
subset of the OSI model that describes the management of the communication
system over the complete seven layers of the OSI reference model, but of
course restricted to the two layers that are defined within a LAN.
-------------------------------------------------------------------------------
*-------------------------------* -----* *------------
] ] ] ]
] AGENT PROCESS ---+-----------------------+-->
] ] ROS(CMIP) ]
] <--+-----------------------+---
] ] ] ]
] A ] ] ] MANAGING
] *--------------+--------* ] ] PROCESS
] ] *-------+--------* -------- ] ]
] ] ] ] V ** ] ] ]
] ]<---->] LME ] LLC ** **] ]LAN ]
] ] ] ] **] OSI ]Station ]
] ] ]-----+----------] Layer 2 ] *------------
] ] ] ] ** ] ]
] ]<---->] LME ] MAC ****] ]
] ] ] ] **] ]
] ] ]-----+----------] -------- ]
] ] ] ] ** ] ]
] ]<---->] LME ] PHY ** **] OSI ]
] ] ] ] **] Layer 1 ]
*-------* *----------------* -----*
** Managed Object
**
-------------------------------------------------------------------------------
Figure 28. A Model for LAN Station Management
An LME or LAYER MANAGEMENT ENTITY exists for each layer of the OSI model as
used by the LAN. An LME provides an interface to the individual layers for
obtaining information about, and effecting changes to, the managed objects
within the layer. Bidirectional communication takes place between the LMEs
and agent process.
As well as the need for the agent process to pass its accumulated data to a
central machine (the managing process), and be able to respond to management
commands, there might also be a need for two Agents to communicate with each
other. For instance, suppose that an established data link between two
applications breaks. Although this might have been caused by a number of
factors, it would be expected that both agent processes, one at each end of
the link, would be notified. The two agent processes on either end of the
data link connection communicate only to exchange a correlator that both use
in subsequent events. Both ends send the reports, and the correlator is used
by the managing process to correlate the two reports. The events from each
side gives a different perspective to the problem and could indicate which
side was causing the problem.
The following figures, Figure 29 under heading "2.4.1.1.2 A Model for LAN
Station Management" and Figure 30 under heading "2.4.1.1.2 A Model for LAN
Station Management" , show these possibilities.
-------------------------------------------------------------------------------
*--------------------*
] Agent Process ] ROS(CMIP) *--------------------*
] ] <---------------> ] Managing Process ]
] *---------------* ] ]
] ] *-------------* ] ]
] ] ]LME ] LLC ] ] ]
] ] ]----+--------] ] ]
] ] ]LME ] MAC ] *--------------------*
] ] ]----+--------]
] ] ]LME ] PHY ]
*----* *-------------*
-------------------------------------------------------------------------------
Figure 29. Information Flow - Agent Process / Managing Process
-------------------------------------------------------------------------------
*--------------------* *--------------------*
] Agent Process ] ROS(CMIP) ] Agent Process ]
] ] <---------------> ] ]
] *---------------* *--------------* ]
] ] *-------------* *------------* ] ]
] ] ]LME ] LLC ] ] LLC ] LME] ] ]
] ] ]----+--------] ]-------+----] ] ]
] ] ]LME ] MAC ] ] MAC ] LME] ] ]
] ] ]----+--------] ]-------+----] ] ]
] ] ]LME ] PHY ] ] PHY ] LME] ] ]
*----* *-------------* *------------* *-----*
-------------------------------------------------------------------------------
Figure 30. Information Flow - Agent Process / Agent Process
2.4.1.2 Registration
In order to allow for the management of a station to begin, it is necessary
for two-way communication between the managing process and the agent process
to take place. The managing process must know that the station, and hence
agent process, is attached to the LAN and is available to be managed. In its
turn, the agent must know to which managing process to send events.
These functions can be provided by a REGISTRATION PROCESS.
NOTE: By showing the protocol flow of this operation, it is also the intent
to help explain the relationship between CMIP and ROS protocols. The relevant
figure is Figure 31 under heading "2.4.1.2 Registration". Also, although OSI
defines the ROS and CMIP protocols, it does not as yet define exactly what,
for instance, an EventType and EventData might be. Therefore, an event, such
as function present, or register request, are implementation dependent.
+------------------------------------------------------------------------------
]
]
]Station Manager. Agent LAN Manager. Management
] process process
]
] -- the Station Manager sends a CMIP EVENT to the Functional
] address of the LAN Manager to announce the process as
] present
]
] -- Note that the CMIP command (EVENT) is enveloped by the ROS INVOKE,
] which is abbreviated to ROIV
]
]
] ROIV(EVENT (Function Present (parameters
] --------------------------------------------->
]
]
] -- If that process is to be managed, then Registration will
] begin
]
] -- The managing process must find a route to the agent, if
] it does not already have one. This can be obtained using
] Find/Found.
]
] FIND
] <---------------------------------------------
] FOUND
] --------------------------------------------->
]
]
] -- Once a route is found, the managing process will send
] a Register request to the Agent. This is sent
] as a CMIP ACTION, imbedded within the ROS INVOKE
]
] ROIV(ACTION (Register Request(parameters
] <---------------------------------------------
]
]
] -- The Agent now responds with a ROS RESULT (RORS)
] showing the results of the ACTION.
]
]
] RORS(ACTION (Register Request(parameters
] --------------------------------------------->
]
+------------------------------------------------------------------------------
Figure 31. The Registration Process
The Agent Processing function is now registered with the Central Managing
Process, until either contact is lost, or a DEREGISTER action is sent.
2.4.2 Resource Management
Due to the fact that the relevant standards bodies have not finished defining
what all the various managed resources actually are, it is difficult to
describe these resources without thinking of what could be managed, or might
need managing in a particular environment. In the token-ring environment, we
can begin to think of the sorts of objects that we know exist, and thus
provide the potential of becoming managed objects. Additionally, if they were
to become managed objects, their usefulness to the management requirements as
stated on page 1 could be seen.
NOTE: It is important to remember that these objects must be considered only
from the architectural or conceptual point of view. By listing them here,
there is no implication that they will ever become the chosen objects, or that
the information about them will become visible to the management process. It
is also important to note that there may be other object classes not listed
here.
Managed objects can be conveniently divided up into a series of groups, which
are known as OBJECT CLASSES . These classes are in fact arranged
hierarchically from a Root, and a sample structure is shown below.
o Root
- Environment
- Managed System
-- Resource Management
-- Token-Ring Layer 2 - MAC
o Counter
o Token-Ring Layer 1
o Name Management
o LAN Layer 2 - LLC
- Counter
- Threshold Pair
o LSAP
- Counter
- Threshold Pair
- LSAP Pair
-- Counter
-- Threshold Pair
o Environment
This object class represents physical equipment, and may for instance
contain information as to the location, machine type and serial number of
the station.
o Managed System
Represents the system as a whole.
o Resource Management
Essentially this equates to a LAN Station Manager application.
o Token-Ring Layer 2
Information here might include the addresses that an adapter will respond
to,(MAC, Functional or Group); the segment (ring) number that it is
attached to, the Upstream neighbor address as well as information on
microcode level.
Counters for this class might include:
- Lost frame errors
- Congestion errors
- Frames aborted during transmit, etc.
o Token-Ring Layer 1
This would represent the physical layer of the token-ring. Included here
could be information on which access unit lobe the adapter is connected
to, which access unit, the speed of the ring, and the wall plug number.
o Name Management
Name Management and the Find/Found process is described in "2.4.3 Name
Management"
o LAN Layer 2 - LLC
Information contained might include the number of SAPs, active and
maximum, the number of active link stations, and the type of LLC services
supported (Type 1,2,3).
Useful counters might be the number of frames and bytes received and
transmitted, or the number of timeouts that have occurred.
A THRESHOLD is simply a relationship between two counters, and a value
used for comparison. Thus, if ERROR and TRAFFIC were counters, a threshold
would be ERROR/TRAFFIC with a comparison of 0.3. This means that if the
ERROR to TRAFFIC Ratio exceeded 0.3, an EVENT should be indicated.
o LSAP
The value of the SAP, the maximum size of the information frame supported,
the number of link stations - all of these might be made available to the
management process, as well as counter values for the number of
connections made to that SAP, bytes and frames transmitted.
o LSAP Pair
Information specific to particular communications would be stored here
including values of window sizes, timers etc. Counters are similar to
those described above for the LSAP and LLC classes.
From the foregoing text, the reader can see that if the managed objects were
chosen correctly, then a wealth of information could be made available to the
management process. Information would be available on machines, their
locations, where they were plugged into the LAN and what their addresses were,
as well as the current condition of the system. There would also be
information available as to the current setting of parameters (for example,
timers) within the system, that control its routine operation. Thus many of
the requirements defined by OSI CONFIGURATION MANAGEMENT would be met.
Some of the counters, for instance the bytes and frames transmitted and
received, might enable determination to be made, not only for some aspects of
PERFORMANCE MANAGEMENT, but for costs, and hence, charge to be made in
accordance with the requirement of ACCOUNTING MANAGEMENT.
FAULT MANAGEMENT would be enhanced beyond the operation of the physical
medium, to include the ability to monitor for intermittent errors that are not
easily visible on the medium. For instance, adapter congestion, badly
formatted messages and timeouts, any of which may be impacting individual
application response times.
SECURITY MANAGEMENT would also be improved, particularly in conjunction with
a Controlled Access Unit, but even without that, the "Registration Process"
and the knowledge that machine identities are logged would enable greater
control of LAN access and assets.
In this discussion, mention has been made of some objects that may need to be
managed, or that may provide useful information to assist with the five areas
OSI defines as requiring management. It must be remembered that the sorts of
objects listed above are only relevant to the token-ring, other type of LAN
would require a different set of objects, maybe even arranged in a different
hierarchy of classes. The IEEE (Institute of Electrical and Electronic
Engineers) are presently defining what these classes and objects may be. With
the advent of FDDI (Fiber Distributed Data Interface), another set of objects
and classes will be required, but this time developed by ANSI (American
National Standards Institute), the body responsible for the standardization of
that technology. It might be reasonable to expect a consistency of approach;
it means however that it may be some time before standard solutions are
available.
2.4.2.1.1 Summary: A LAN Station Manager is an application that may reside
in any station on a LAN. It:
o Is an application that maps to the concept of an AGENT PROCESS as defined
by OSI Systems Management
o Maintains managed objects, that themselves maybe counters and pertinent
configuration information about LAN attachments
o Responds to queries regarding the above information from the central
managing process
o Allows the managing process to set certain configuration and threshold
values
o Allows for one or more managing processes for a single agent process
o Reports events to that process.
Also, if using OSI techniques, A LAN Station Manager can provide:
o A standardized set of information, which enables much greater visibility
of the functional areas described by Open Systems Management.
o A standardized model for gathering and controlling the information.
o A standardized protocol for communicating with a central management
application.
o It allows the network and systems management protocols to be independent
of the "transport pipes" through which they flow.
All this means that the ability to provide a central LAN Management process is
enabled, and to provide a consistent set of information from LAN-attached
stations whatever their make or type, to the benefit of the organization
running the LAN.
A subset of the concept of LAN Station Manager, the agent process, is
implemented in the:
o IBM LAN Station Manager Version 1.0 - see "9.5 IBM LAN Station Manager
V1.0."
o IBM 8230 Controlled Access Unit - see "5.1.1.2 IBM 8230 Controlled Access
Unit."
The LAN Managing process is implemented by the IBM LAN Network Manager V1.0,
IBM LAN Network Manager V1.1, and IBM LAN Network Manager Entry V1.0 - see
"9.3.5 IBM LAN Network Manager V1."
2.4.3 Name Management
A Find/Found function has been implemented in IBM LAN Network Manager, IBM LAN
Network Manager V1.1, IBM LAN Network Manager Entry V1.0 and IBM 8230
Controlled Access Unit to simplify the process whereby the Managing process
registers the LAN Stations and 8230's. It allows an application name to be
registered on a workstation. Then using the Find/Found function LAN
workstations can 'find' each other and determine a route with which to
communicate.
For instance, the IBM LAN Network Manager can dynamically 'find' an 8230 that
it is responsible for and register it.
3.0 LAN Architectures and Standards
In the previous chapter, "2.0 LAN Technology," various technological
approaches were described for both physical LAN attachment considerations and
medium access methods. Some alternative approaches to LAN interconnection
were also introduced.
The 802 Project of the Institute of Electrical and Electronics Engineers
(IEEE) has produced a set of standards for LAN architectures, giving vendors
guidance for producing LAN products and users a choice of standardized local
area networks with a certain degree of interconnectivity. The IEEE LAN
standards align with the bottom two layers of the International Standards
Organization's (ISO) model for Open Systems Interconnection, referred to as
the OSI REFERENCE MODEL.
3.1 LAN and Communications Standards
The International Standards Organization (ISO) developed the seven-layer OSI
Reference Model to provide a standard reference for intercommunication between
computer systems through a network using common protocols.
The ISO Reference Model, depicted in Figure 33 under heading "3.1.2 OSI and
SNA Models", became an international standard in 1983 (IS 7498). Each layer
addresses a well-defined section of the total architecture. The Layers of the
OSI Reference Model are, from top to bottom:
o Application Layer
The application layer gives the user access to all the lower OSI
functions. The purpose of this layer is to support semantic exchanges
between applications existing in "open" systems.
o Presentation Layer
The presentation layer is concerned with the representation of user or
system data. This includes necessary conversions (for example, printer
control characters) and code translation (for example, ASCII to or from
EBCDIC).
o Session Layer
The session layer provides mechanisms for organizing and structuring
interaction between applications and/or devices.
o Transport Layer
The transport layer provides transparent and reliable end-to-end data
transfer, relying on lower layer functions for handling the peculiarities
of the actual transfer medium.
o Network Layer
The network layer provides the means to establish connections between
networks. The standard also includes procedures for the operational
control of inter-network communications and for the routing of information
through multiple networks.
o Data Link Layer
The data link layer provides the functions and protocols to transfer data
between network entities and to detect (and possibly correct) errors that
may occur in the physical layer.
o Physical Layer
The physical layer is responsible for physically transmitting the data
over the communications link. It provides the mechanical, electrical,
functional and procedural standards to access the physical medium.
This layered approach was selected as a basis for the OSI Reference Model to
provide flexibility and open-ended capability through defined interfaces. The
interfaces permit some layers to be changed while leaving other layers
unchanged. In principle, as long as standard interfaces to the adjacent
layers are adhered to, an implementation can still work. For example, a
system implementation could use either HDLC or local area network protocols as
the data link layer. Similarly, a particular layer such as the presentation
layer, can be implemented as a null layer for the time being. This means the
layer is functionally empty, providing only the mandatory interfaces between
the upper and lower layers (application and session layers respectively). The
Manufacturing Automation Protocols (MAP 2.1) "3.2.1.2 IEEE 802.4, ISO 8802-4")
exemplify implementation with a null layer.
3.1.1 IEEE 802 and OSI
In February 1980, The Institute of Electrical and Electronic Engineers' (IEEE)
Computer Society established "Project 802" to draft standards for local area
networks. In keeping with the OSI approach, IEEE Project 802 created a
reference model with two layers (which correspond to the data link and
physical layers of the OSI model). In the IEEE model, however, the data link
layer is further divided into two sublayers: the LOGICAL LINK CONTROL (LLC)
sublayer, and the MEDIUM ACCESS CONTROL (MAC) sublayer.
Details on these sublayers can be found in "3.2 LAN Standards."
Due to the variety of technically sound approaches proposed for local area
networks, IEEE Project 802 decided to draft standards for the most probable
implementations:
o CSMA/CD (Contention Bus)
o Token-passing bus
o Token-passing ring.
The commonly used names for these standards are derived from the project's
initial designation "802". Hence, we have:
o IEEE 802.1 - Higher level interface standard
o IEEE 802.2 - Logical link control standard
o IEEE 802.3 - CSMA/CD standard
o IEEE 802.4 - Token-passing bus standard
o IEEE 802.5 - Token-passing ring standard.
Over the years, the IEEE work has expanded, with the formation of new
subcommittees. These include:
o IEEE 802.6 - Metropolitan Area Networks (MANs)
o IEEE 802.7 - Broadband Technical Advisory Group
o IEEE 802.8 - Fiber Technical Advisory Group
o IEEE 802.9 - Integrated Voice/Data on LAN
o IEEE 802.10 - Interoperable LAN Security
o IEEE 802.11 - Wire-less LANs.
The IEEE 802.1 Higher Level Interface subcommittee is currently finalizing the
draft IEEE 802.1 standard, also presented to the ISO as an ISO draft proposal.
This draft includes the following subjects:
o IEEE 802.1 Part A which describes the relationship of the IEEE 802 work to
the ISO Open Systems Interconnection Basic Reference Model.
o IEEE 802.1 Part B which specifies an architecture and protocol for the
management of IEEE 802 LANs.
o IEEE 802.1 Part D which specifies an architecture and protocol for the
interconnection of IEEE 802 LANs below the MAC service boundary.
o IEEE 802.1 Part E which deals with a Station Load Protocol, used for IPL
of workstations from a server.
o IEEE 802.1 Part F which addresses guidelines for the development of layer
management standards and definition of managed objects.
Figure 32 under heading "3.1.1 IEEE 802 and OSI" depicts the relationship
between the IEEE 802.x projects and standards.
+------------------------------------------------------------------------------
]
]
] To Layer
] A 7
] ]
]] ] ] *-----------------------------------------------------------------
]] ] ]
]] IEEE ] ] IEEE 802.1
]] 802.10 ] ] Higher-Level Interface and Interworking
]] Inter- ] ]
]]Operable] ] *------------------------------------------------------------
]] LAN ] ] ]*-----------------------------------------------------------
]]Security] ] ]]
]] ] ] ]]
]] ] ] ]] IEEE 802.2
]] ] ] ]] Logical Link Control
]] ] ] ]]
]] ] ] ]]
]*--------* ] ]*-----------------------------------------------------------
] ] ]*----------* *----------* *----------* *----------* *-------
] ] ]]IEEE 802.3] ]IEEE 802.4] ]IEEE 802.5] ]IEEE 802.6] ]IEEE 80
] ] ]] CSMA/CD ] ]Token Bus ] ]Token-Ring] ] ] ]
] ] ]] Medium ] ] Medium ] ] Medium ] ] ] ]
] ] ]] Access ] ] Access ] ] Access ] ] Metro- ] ]Integra
] ] ]]----------] ]----------] ]----------] ] politan ] ] Voice
] ] ]] ] ] ] ]Token-Ring] ] Area ] ] Data
] ] ]] CSMA/CD ] ]Token Bus ] ] Media ] ] ] ]
] ] ]] Media ] ] Media ] ]IBM Cablng] ] ] ]
] ] ]] ] ] ] ] System ] ] ] ]
] *----**----------* *----------* *----------* *----------* *-------
]
]
] *------------------* *------------------*
] ] IEEE 802.7 ] ] IEEE 802.8 ]
] ] Broadband ] ] Fiber Optic ]
] ] Tech. Adv. Group ] ] Tech. Adv. Group ]
] *------------------* *------------------*
]
]
]
]
+------------------------------------------------------------------------------
Figure 32. Relationships Among the 802 Standards
The International Standards Organization has since adopted the IEEE 802
standards as part of the OSI Reference Model, and has given them the following
ISO numbers:
o IEEE 802.2 Currently an OSI International Standard, IS 8802-2 dated 1989
o IEEE 802.3 Currently an OSI International Standard, IS 8802-3 dated 1989
o IEEE 802.4 Currently an OSI International Standard, IS 8802-4 currently
unpublished
o IEEE 802.5 Currently an OSI International Standard, IS 8802-5. currently
unpublished
Before publishing 8802-5, ISO was waiting for the result of balloting on four
Proposed Draft Addenda (PDADs) to the standard. These were:
o PDAD 1 - Token ring operation at 16Mbps
o PDAD 2 - MAC sublayer enhancements
o PDAD 3 - Station management entity specification
o PDAD 4 - Source routing.
As of this time, PDADs 1, 2, and 3 have all progressed within ISO to Draft
Addenda, which will be lumped together with 8802-5, and a new Draft
International Standard DIS 8802-5.2 will be voted on and then published.
PDAD 4 on Source Routing has recently and dramatically changed shape. Due to
the advent of the Source Routing Transparent (SRT) Bridge, described in "6.3.3
Source Routing/Transparent Bridging Inter-operability," the addendum will now
only contain details of token-ring frame formats including the Routing
Information field. All the rest of the work has moved into IEEE 802.1 Part D.
It is hoped that the balloting on PDAD 4 will be finished in time to be
incorporated in the DIS ballot for 8802-5.2.
3.1.2 OSI and SNA Models
IBM System Network Architecture (SNA) is a proprietary layered architecture
which, although defined earlier than the OSI Reference Model, is based upon a
seven-layer structure. SNA was announced as IBM's networking architecture in
1974, and the first products implementing the SNA protocols became available
in the mid 1970s. Since then, SNA has continued to grow and evolve. Part of
this evolution has included IBM's support of standard communications
interfaces through continued publication and update of SNA specifications, and
through development of enhancements to SNA closely coordinated with emerging
international standards. In this way IBM developed the IBM Token-Ring Network
as an enhancement to SNA while participating actively in the development of
the IEEE 802.5 and 802.2 standards.
Figure 33 under heading "3.1.2 OSI and SNA Models" shows a comparison of the
OSI, IEEE 802 and the SNA models.
+------------------------------------------------------------------------------
]
]
] OSI SNA
] Model Layers
] *------------* *-----------------
] ]Application ] ] Transaction
] ]------------] ]-----------------
] ]Presentation] ] Presentation
] ]------------] ]-----------------
] ] Session ] ]Data Flow Control
] ]------------] ]-----------------
] ] Transport ] ]Transmission Ctrl
] ]------------] ]-----------------
] ] Network ] IEEE 802 Model ] Path Control
] ]------------] *----------------------------* ]-----------------
] ] Data ] *--> ]Logical Link Control] ] ] Data
] ] ] <--] ]--------------------]Station] ] Link
] ] Link ] *--> ]Medium Access Ctrl ] ] ] Control
] ]------------] ]--------------------] Mgt. ] ]-----------------
] ] Physical ] <-----> ] Physical Control ] ] ] Physical
] *------------* *----------------------------* *-----------------
]
]
+------------------------------------------------------------------------------
Figure 33. OSI Reference Model, IEEE 802 Local Area Network Model, and SNA Mode
Although there is general correspondence between the OSI and the SNA layers,
the detailed protocols and services vary. The SNA layers (from the top down)
are:
o Transaction services
The transaction services layer provides application services to the end
user of an SNA network. These services are also used by IBM Office
Architectures: Document Interchange Architecture (DIA) for document
distribution between office systems, and SNA Distribution Services (SNADS)
for asynchronous data distribution between distributed applications and
office systems.
The transaction services layer also provides configuration, session and
management services to control network operation.
o Presentation services
The presentation services layer is concerned with the representation of
application, end-user, and system data. The services provided include
3270 data stream support and intelligent printer data stream support.
The presentation services layer also defines protocols for
program-to-program communication, and controls conversation-level
communication between transaction programs.
o Data flow control
The data flow control layer provides flow control services for logical
unit to logical unit (LU-LU) sessions. It does this by assigning sequence
numbers, correlating requests and responses, enforcing session request and
response mode protocols and coordinating session send and receive modes.
o Transmission control
The transmission control layer provides basic control of the transmission
resources in the network, verifying received sequence numbers, and
managing session level pacing. Boundary function support for peripheral
nodes is provided by transmission control, and data encryption can also be
supported by this layer's services.
o Path control
The path control layer provides protocols to route message units through
an SNA network, and to interconnected SNA networks (via System Network
Interconnect - SNI).
o Data link control
The data link control layer provides protocols for transferring message
units across the physical link between two nodes, and also provides
link-level flow control and error recovery. The data link control layer
supports SDLC, System/370 data channel, X.25 and IEEE 802.2 and 802.5
protocols.
o Physical control
The physical control layer provides a physical interface for any
transmission medium attached to it. This layer defines the signaling
characteristics needed to establish, maintain, and terminate physical
connections for supported attachments.
3.2 LAN Standards
The data link layer provides the protocols used to physically transmit data on
a communications link. These protocols must be able to detect when a
transmission has been corrupted by errors and to retransmit the data.
The IEEE 802 project divided the data link layer into two sublayers, the
MEDIUM ACCESS CONTROL (MAC) layer and the LOGICAL LINK CONTROL (LLC) layer.
These sublayers are discussed in the following sections.
Today, there is a growing requirement to further divide the physical layer
into a PHYSICAL and a PHYSICAL MEDIUM DEPENDENT layer. This approach was seen
in the discussion on FDDI. See "2.3 Fiber Distributed Data Interface
Concepts."
As previously mentioned, the IEEE 802.1 subcommittee work has not yet been
completed, especially in the areas of the management and interconnection of
IEEE 802 LANs.
3.2.1 Physical Layer and MAC Sublayer
The medium access control (MAC) sublayer contains mechanisms to control
transmissions on the LAN so that two or more stations don't try to transmit
data at the same time, logic to control whether a station on the LAN is in
transmit, repeat, or receive state, and addressing schemes to control the
routing of data on the LAN. It also provides some or all of the following
functions:
o MAC addressing. A MAC address is the physical address of the station
(that is, device adapter) on the LAN, a pre-defined group address which
is recognized by adapters of devices belonging to the group, a broadcast
address which is recognized by all adapters on the LAN or a null address
for frames which should not be received by any station on the LAN. The
MAC address is used to identify the physical destination and source of
anything transmitted on the LAN.
o Frame copying. This is the "receipt", that is, copying of a frame into
the buffers of an attached adapter which recognizes its own address in the
destination address field of a frame.
o Frame type recognition. This is the identification of the type (for
example, system or user) and format of a frame of data.
o Frame control. This function ensures that frames can be processed
accurately by providing frame check sequence numbers and starting or
ending frame delimiters.
o Priority management. This function allows preferential access to the
medium while maintaining fairness with respect to all participating
stations.
o LAN management. A collection of protocols has been defined to support
monitoring of a LAN and the ability to handle error conditions at the
access control level.
Some access protocols do not provide priority management or frame control
functions, but rely on higher layer services to provide these functions.
3.2.1.1 IEEE 802.3, IS 8802-3
The IEEE 802.3 standard describes the access method and the physical layer
specifications for a contention bus LAN. The standard specifies a
1-persistent carrier sense multiple access with collision detection (CSMA/CD)
access protocol, using baseband transmission on a single channel pair at
10Mbps. Binary data is transmitted as signal elements on a common shielded 50
Ohm coax cable using the Manchester code scheme. Baseband bus is the most
popular IEEE 802.3 topology, although a star-wired baseband bus and broadband
tree are also part of the standard.
The maximum allowable non-repeated distance between stations is a 500-meter
segment. One or more segments may attach to a particular segment using
repeaters. However the maximum number of segments in any end-to-end path is
five, that is, a maximum of four repeaters in any path. Within any path, no
more than three segments may have stations attached to them, while the maximum
number of attachments to one segment is 100.
For the total LAN, 1024 attachments are allowed, with a total end-to-end
distance between stations not exceeding 2.5 kilometers.
The standard also describes the TRANSCEIVER UNIT, used to couple the Data
Terminal Equipment (DTE) to the medium. The transceiver unit components
together with the associated medium attachment taps cause signal reflections
and standing waves (second and third harmonic distortion) due to bridging
impedances whenever a transmission occurs. The unusual 50 Ohm coaxial cable
was selected to limit the capacitance caused by the taps. Because of these
reflections and standing waves, the placement of transceivers along the cable
and the lengths of the cables themselves must be controlled to ensure that
transmissions are not disrupted. For example, if the standing waves set up by
one station's transmission were strong enough, they could be mistaken by that
transmitting station for another transmitting station; that is, the
transmitting station would erroneously detect a collision. Station attachment
to the common bus must be separated by multiples of 2.5 meters to prevent
reflections caused by the taps from adding up in phase.
IS 8802-3 contains standards for several different implementations of CSMA/CD
LANs. These include:
o 10base5 - 10Mbps, baseband, maximum bus length of a segment is 500 meters
o 10base2 - 10Mbps, baseband, maximum bus length of a segment is 200 meters.
Additional work is progressing on:
o 10broad36 - an IEEE standard for a 10Mbps, broadband network, but not yet
included in ISO
o 10baseT - CSMA/CD LAN on Telephone Twisted Pair wiring
o 10baseF - a CSMA/CD implementation on Optical Fiber.
An early IEEE standard, 1base5, implemented commercially as Starlan(**) was
not incorporated into 8802-3.
3.2.1.1.1 Service Primitives: The basic MAC service primitives used in this
and all IEEE MAC standards are:
o Medium access data request (MA_DATA.request)
This primitive is generated whenever the LLC sublayer has data to be
transmitted to another station(s) on the LAN. The MAC sublayer formats it
in a MAC frame and transmits it.
o Medium access data confirm (MA_DATA.confirm)
This primitive is generated by the MAC sublayer in response to a
"MA_DATA.request" from the local LLC sublayer. A status parameter is used
to indicate the outcome of the associated MA_DATA.request.
o Medium access data indicate (MA_DATA.indicate)
This primitive is sent to indicate that a valid frame arrived at the local
MAC layer. The frame was transmitted without errors and was correctly
addressed.
o Medium access data response (MA_DATA.response)
This primitive is used as a response to a MA_DATA.indicate.
One use of these primitives is shown in Figure 34 under heading "3.2.1.1.1
Service Primitives". The figure shows two stations, A and B, with the LLC
layer of station A requesting transmission of a frame to the MAC service
interface. Upon receipt of the frame by station B, an indicate is generated,
notifying the LLC of an incoming frame. LLC generates response to indicate
that it has the frame. The confirm indicates to station A that the frame was
transmitted without error.
+------------------------------------------------------------------------------
]
] Station A Station B
] MAC MAC
] Service I/F Service I/F
] ] ]
] ] ]
] MA_DATA.request ] ]
] ---------------->] ]
] ] ]MA_DATA.indicate
] ] ]--------------->
] ] ]
] ] LAN ]
] ] ]MA_DATA.response
] ] ]<---------------
] ] ]
] MA_DATA.confirm ] ]
] <----------------] ]
] ] ]
+------------------------------------------------------------------------------
Figure 34. IEEE MAC Primitives
The CSMA/CD medium access protocol was described in "2.2.1 Basic CSMA/CD
Concepts." Figure 9 under heading "2.2.1.1.1 Frame Format" shows the
eight-field structure of the 802.3 standard MAC frame.
3.2.1.1.2 Summary: The IEEE 802.3 standard is a contention-based protocol.
For low traffic volumes and relatively short messages, an IEEE 802.3 LAN can
provide the best response time. However, because of the instability
introduced by collision recovery during periods of high utilization, response
times and performance cannot be predicted reliably. At the MAC level, the
IEEE 802.3 protocol cannot guarantee access to the medium. These performance
considerations make IEEE 802.3 protocol LANs less desirable than other
protocols for backbone LANs.
All stations have equal rights to transmit on an idle bus. In low-traffic
situations, this minimizes the need to handle priority requests. However,
with heavier traffic, the lack of priority support within the protocol may be
a problem for some applications.
Special considerations may be required for applications or data with security
requirements, because IEEE 802.3 stations broadcast their frames
simultaneously to all other LAN stations on the network which therefore can
receive and copy any transmitted data.
Network management is not defined within the IEEE 802.3 standard. It is under
consideration as part of the IEEE 802.1 sub-project (see "3.1.1 IEEE 802 and
OSI").
3.2.1.2 IEEE 802.4, ISO 8802-4
The IEEE 802.4 standard describes the token-passing bus access protocol and
its physical layer specifications. In this type of LAN, stations on the
network are physically connected to a bus, but access to the bus is controlled
as if it were a logical ring. Each station keeps track of the addresses of
its logical neighbors (that is, those which immediately precede and follow it
in a logical sequence). The physical connection sequence on the bus is
independent from the order of the logical ring.
A token is used to control access to the bus. Once a station has control of a
token, it has complete control of the bus for a defined period of time (the
token_holding time). In addition to transmitting one or more frames, the
controlling station can poll other stations and receive responses or
acknowledgements during this time period. When it wishes to relinquish control
of the bus or when the token_holding timer expires, the station passes the
token on to its successor station.
The IEEE 802.4 standard defines broadband transmission on shielded coax cable
at data rates of 1, 5 or 10 Mbps. It provides a choice of different
modulation techniques and cable types. There are three standard topologies,
each involving different modulation techniques and cable types used for the
trunk and drops. The three topologies have the following characteristics:
o Omni-directional bus at 1 Mbps, using Manchester encoding.
o Omni-directional bus at 5 or 10 Mbps.
o Directional bus with active head-end repeater at 1, 5 or 10 Mbps.
The token-passing bus medium access protocol was described earlier in "2.2.3
Basic Token-Passing Bus Concepts."
In summary, one or more stations on the bus must perform the following
functions:
o RING INITIALIZATION, performed when the network is first powered up and
after a catastrophic error.
o STATION ADDITION, optionally performed when a station holding the token
accepts the insertion of a new successor station (that is, a new station
with an address that is between that of the station holding the token and
its current successor station.
o STATION REMOVAL, achieved by sending a new successor identification to its
predecessor, or by just disconnecting from the LAN. In the latter case,
recovery mechanisms will establish the proper new logical ring
configuration.
o RECOVERY AND MANAGEMENT, including recovery from the following types of
failures:
- Bus idle (lack of activity on the bus).
- Token-passing failure (lack of valid frame transmission).
- Duplicate token (detected by the token-holding station).
- Detection of a station with a faulty receiver.
- Duplicate MAC addresses.
Figure 21 under heading "2.2.3.1.1 Frame Format" shows the generic 802.4
standard MAC frame format as well as the IEEE 802.4 token format which is
transmitted as a special frame.
3.2.1.2.1 IBM and IEEE 802.4 - ISO IS 8802-4: In 1984, IBM issued a
statement of direction stating IBM's intention to implement the Manufacturing
Automation Protocol (MAP) standard based upon approval of the MAP Version 3
specifications. IBM also refers to MAP as the INDUSTRIAL LAN, which is part
of the IBM COMPUTER-INTEGRATED MANUFACTURING (CIM) product offerings. The
ultimate goal of CIM is integration of all processes involved in a
manufacturing environment. Integration is based on controlling the information
flow for which MAP as a full seven-layer communications architecture will
provide networking and applications support. The ISO 8802-4 standard provides
the two lowest layers of this seven-layer architecture.
3.2.1.2.2 Manufacturing Automation Protocol: Work on a MAP standard started
in 1980 as a task force within the General Motors Corporation. Gradually
other manufacturers, including IBM, began to participate in the definition of
the protocols and the implementation of products.
The following milestones have occurred in the development of the MAP standard:
o MAP 1.0 was released in April 1984 as the result of the General Motors
task force.
o Based upon the participation of additional companies in the
standardization process, MAP 2.1 was issued in March 1985 as a set of
interim specifications to allow early product development and experience.
o In December 1985, MAP 2.1 was updated to correct some mistakes in the
earlier release.
o MAP 2.2 was released in March 1987, with the addition of carrierband
network specifications.
o The ultimate standard, referred to as MAP 3.0, was released as a draft in
April 1987. Final release occurred in 1988.
Figure 35 under heading "3.2.1.2.2 Manufacturing Automation Protocol" shows
the relationship between the seven-layer OSI Reference Model and MAP 2.1.
+------------------------------------------------------------------------------
]
]
] OSI Layers MAP 2.1
]
] *---------------------------*
] ] ] MMFS/EIA 1393A
] ] 7 Application ] ISO CASE Kernel
] ] ] ISO FTAM Subset
] ] ] Directory Services Subset
] ]---------------------------]
] ] 6 Presentation ] null layer
] ] ]
] ]---------------------------]
] ] 5 Session ] ISO Session Kernel 8327
] ] ]
] ]---------------------------]
] ] 4 Transport ] ISO Transport (Class 4) 8073
] ] ]
] ]---------------------------]
] ] 3 Network ] ISO Connectionless Internet
] ] ] Protocol 8473
] ]---------------------------]
] ] 2 Data ] Logical Link Ctl ] ISO 8802-2 Link Level (Class I)
] ] ] - - - - - - - - -]
] ] link ] medium access ctl] Token-passing bus
] ]---------------------------] ISO 8802-4
] ] 1 Physical ] Broadband Media
] ] ]
] *---------------------------*
]
]
]
+------------------------------------------------------------------------------
Figure 35. MAP 2.1 and the OSI Reference Model
MAP 2.1 specifications have been subject to two main criticisms:
1. When multi-channel capability is not required, the full broadband
implementation is excessively expensive and not required. Therefore some
potential users and vendors preferred a carrierband LAN implementation.
2. The processing of a full seven-layer implementation of the architecture
may require too much processing time for smaller systems in time-critical
applications. Initially, this critique was addressed by two different
approaches, MAP/ENHANCED PERFORMANCE ARCHITECTURE (MAP/EPA) and MINIMAP
(see Figure 36 under heading "3.2.1.2.2 Manufacturing Automation
Protocol").
+------------------------------------------------------------------------------
]
]
] MAP/Enhanced Performance
] Architecture
]
] *--------* *-------------*
] ]MAP/OSI ] ]Time-Critical]
] ]Applic. ]<----->] Application ]<--*
] *--------* *-------------* ]
] ] ] ] MiniMAP
] *------------------* ]
] ] 7 ] I ] ] *------------------*
] ]--------] n ] ] ] Time-Critical ]
] ] 6 ] t ] *--->] Application ]
] ]--------] e ] *------------------*
] ] 5 ] r ] ]
] ]--------] f ] *------------------*
] ] 4 ] a ] ] Interface ]
] ]--------] c ] *------------------*
] ] 3 ] e ] ]
] ]------------------] *------------------*
] ] ] LLC ] ] ] LLC ]
] ] 2 ] - - - - - -] ] 2 ] - - - - - -]
] ] ] MAC ] ] ] MAC ]
] ]------------------] ]------------------]
] ] 1 Physical ] ] 1 Physical ]
] *------------------* *------------------*
] ] ]
] ] ]
] ------------------- -------------------
] Physical Media Physical Media
]
]
+------------------------------------------------------------------------------
Figure 36. MAP/EPA and MiniMAP
MAP Version 3 enhances MAP 2.1 by:
o Imbedding the MAP/EPA and MiniMAP approaches into the standard to meet the
time-dependent requirements of the process industry.
o Replacing MMFS by the Manufacturing Messaging Specifications (MMS).
o Including OSI presentation layer 6 support.
o Extending layer 4 class 4 service dealing with error detection and
recovery.
With respect to network interconnection, the MAP approach is to provide a
full, seven-layer MAP 3.0 backbone configuration at the plant floor
interconnected with other carrierband MAP LANs and other standard LANs such as
as IEEE 802.3 and IEEE 802.5. This approach addresses the objective to
integrate technical design, office and business applications, computer-aided
production management and manufacturing with the communications requirements
of the individual production cells as shown in a sample configuration in
Figure 37 under heading "3.2.1.2.2 Manufacturing Automation Protocol".
+------------------------------------------------------------------------------
]
]
]
] *------------ Technical Design
] ]
] *------------------*
] X.25<---* *---* ] ]-- Office appl.
] *---]G/W]--] Token-Ring LAN ]
] Info.<---* *------] ]-- Business appl.
] Provider *---]G/W] *------------------*
] *---* ]
] *-------*
] ]Bridge/]
] ]Router ]
] *-------*
] ]
] Computer-Aided Production Management and Manufacture
] Full MAP ] Other 8802-2 LAN
] *---* *---* *---* ] *---* *---* *---*
] *---* *---* *---* ] *---* *---* *---*
] *-----+-----* ] *-----+-----*
] *-------* ] *-------*
] ]Router ] ] ]Bridge ]
] *-------* ] *-------*
] <-------------------------------------------------------------> Full MAP
] <-------------------------------------------------------------> Broadband
] ] *---* backbone
] *-------* ]G/W]
] ]Bridge/] *---*
] ]Router ] ]
] *-------* **
] ] ]
] ] X.25 to remote sites
] inter- and intra-cell communications
] <-------------------------------------------------------------> Baseband
] <-------------------------------------------------------------> MAP
] *-------* *-------*
] ]Gateway] ]MAP/EPA]
] *-------* *-------*
] *-----+-----* *-----+-----*
] *---* *---* *---* Production *---* *---* *---* Production
] *---* *---* *---* cell *---* *---* *---* cell
] Proprietary N/W MiniMAP
]
]
+------------------------------------------------------------------------------
Figure 37. MAP Integrated Network Configuration
An important concern which appeared during the evolution of the Manufacturing
Automation Protocol is associated with migration from MAP 2.1 local area
networks to a MAP 3.0 environment. This may be difficult because of the
nature of the changes. Additional information is provided in "5.4.1 The 8232
LAN Channel Station and Industrial LAN."
3.2.1.2.3 Summary: The token-passing bus access method is predominately used
for industrial LANs. It uses a logical ring on a physical bus, and has a
number of transmission techniques and data rates to choose from.
The protocol optionally provides eight priority service classes to higher
level data frames, handled at the MAC level by four access classes based upon
the current traffic load on the bus.
With respect to a full seven-layer architecture for the manufacturing
environment, application implementation is very limited because of the
immaturity of the standards, but will likely increase as the standards are
finalized.
3.2.1.3 IEEE 802.5, ISO IS 8802-5
The IEEE 802.5 standard describes the token-passing ring medium access
protocol and its physical attachments.
In a token-passing ring network the stations on the LAN are physically
connected to a wiring concentrator usually in a star-wired ring topology.
Logically, stations are connected in a pure ring topology. Each station has
driver/transmitter as well as receiver circuitry (see Figure 38 under heading
"3.2.1.3 IEEE 802.5, ISO IS 8802-5").
Differential Manchester code is used to convert binary data into signal
elements, which are transmitted at 1 or 4 Mbps IEEE standard speeds. An
update to the standards to support 16 Mbps is currently in the process of
review and approval. The standard does not prescribe the type of cabling to
be used. In IBM's Token-Ring Network implementation, shielded twisted pair
cabling is recommended.
+------------------------------------------------------------------------------
]
]
]
] *-----* *-----*
] ] S2 ] ] S3 ]
] ]-----] ]-----]
] *-----]R ] D]-------]R ] D]-----*
] ] *-----* *-----* ]
] *----* *---------> *----*
] ] ]D] ] Token ]R] ]
] ]S1]-] ] ]-]S4]
] ] ]R] ] ]D] ]
] *----* *--- *----*
] ] *-----* *-----* ]
] *-----]D ] R]-------]D ] R]-----*
] ]-----] ]-----]
] ] S6 ] ] S5 ]
] *-----* *-----*
]
]
] D = driver/transmitter
] R = Receiver
] S1 to S6 are ring stations
]
+------------------------------------------------------------------------------
Figure 38. Sample Ring Configuration
Access to the ring is controlled by a circulating token. A station with data
to transmit waits for a free token to arrive. When a token arrives, the
station changes the token into a frame, appends data to it and transmits the
frame. If the destination station is active, it will copy the frame and set
the "frame copied" and "address recognized" bits, providing MAC level
acknowledgement to the transmitting station. The sending station must strip
the frame from the ring and release a new token onto the ring.
An option in the architecture allows the sending station to release a token
immediately after transmitting the frame trailer, whether or not the frame
header information has already returned. This is called EARLY_TOKEN RELEASE
and tends to reduce the amount of idle time in higher speed token-passing
rings running at, for example, 16 Mbps.
The token-passing protocol provides an extensive set of inherent fault
isolation and error recovery function, for implementation in every attaching
device. The adapter network management functions include:
o Power-on and ring insertion diagnostics.
o Lobe-insertion testing and online lobe fault detection.
o Signal loss detection, beacon support for automatic test and removal.
o Active and standby monitor functions.
o Ring transmission errors detection and reporting.
o Failing components isolation for automatic or manual recovery.
The token-passing ring medium access protocol was described earlier in greater
detail (see "2.2.2 Basic Token-Passing Ring Concepts").
In summary, the token-passing ring protocol is based on the following
cornerstones:
o ACTIVE MONITOR
- Ensures proper ring delay.
- Triggers Neighbor_Notification.
- Monitors token and frame transmission.
- Detects lost tokens and frames.
- Purges circulating tokens or frames from the ring.
- Performs auto-removal in case of multiple active monitors.
o STANDBY MONITOR (any other ring station)
Detects failures in the active monitor and disruptions on the ring.
o TOKEN_CLAIMING PROCESS
By which a new active monitor is elected when the current active monitor
fails. This process may be initiated by the current active monitor or by a
standby monitor.
Figure 16 under heading "2.2.2.1.1 Data Transmission" shows the format of a
802.5 standard MAC frame as well as the token format and the format of the
abort delimiter.
The architecture describes 28 different MAC control frames, each identified by
a unique Major Vector Identifier (MVID). The main ones have been described in
"2.2.2 Basic Token-Passing Ring Concepts" and are referred to as:
o Active Monitor Present MAC frame.
o Ring Purge MAC frame.
o Standby Monitor Present MAC frame.
o Claim Token MAC frame.
o Lobe Media Test MAC frame.
o Duplicate Address Test MAC frame.
o Request Initialization MAC frame.
o Beacon MAC frame.
o Soft Error Report MAC frame.
3.2.1.3.1 Summary: The token-passing protocol provides for efficient use of
the media under both light and heavy traffic loads. It guarantees fair access
to all participating stations. This fairness is enhanced by an eight-level
priority mechanism, based on priority reservations made in a passing token or
frame. A key benefit of the token-passing ring protocol is its ability to
handle increased traffic loads or peaks, making it an ideal protocol for
larger and/or more heavily used LANs (including backbone rings). This also
makes it a good base LAN for connection to even higher bandwidth LANs such as
FDDI.
3.2.1.4 Fiber Distributed Data Interface
The AMERICAN NATIONAL STANDARDS INSTITUTE (ANSI) standards working group
X3T9.5 has formulated a draft proposal for an international standard called
FIBER DISTRIBUTED DATA INTERFACE (FDDI).
A detailed description of the FDDI LAN was given in "2.3 Fiber Distributed
Data Interface Concepts."
In terms of current standards, FDDI is described by ISO 9314, which is
subdivided into 6 parts. These are:
o 9314-1 - the physical layer
o 9314-2 - the MAC layer
o 9314-3 - the physical medium dependent layer
o 9314-4 - use of FDDI on single-mode fiber
o 9314-5 - Hybrid Ring Control - FDDI II, which divides the FDDI frame into
slots, such that frames may contain both synchronous and asynchronous
data. This implementation is more suitable for voice traffic.
o Station Management has as yet not been allocated a 9314 number.
At this time, not all of these are International Standards. Further
information can be found in "2.3.1.2 FDDI Standards."
3.2.2 Logical Link Control Sublayer
The IEEE 802.2 standard describes the top sublayer of the data link layer. It
is common to all the MAC sublayers layers defined by the IEEE. See Figure 32
under heading "3.1.1 IEEE 802 and OSI". This means higher layer protocols are
shielded from the peculiarities of the physical medium as well as from the
specific medium access protocol being used. Obviously, limitations such as
throughput characteristics and number of attachments, inherent in each of the
lower level standard protocols, apply equally to higher layer protocols.
Local area networks (as well as wide area networks) need the function of a DLC
layer to optimize the capacity, accuracy, and availability of the physical
medium.
The function of a network layer, to route data through the network and set up
a connection between two end points, is also needed in a LAN. Basic LAN
routing, error control, and flow control is defined as part of the LLC
sublayer while the network layer may be a null layer.
In summary, logical link control provides a consistent view of a LAN to the
upper layers regardless of the media and protocols being used and may support
all required end-to-end connectivity services.
The interface to the upper layers is provided through LLC SERVICE ACCESS
POINTS (LSAP), just as Service Access Points (SAPs) are the architected
interfaces between any two adjacent layers in the OSI Reference Model
Figure 33 under heading "3.1.2 OSI and SNA Models". A station can have more
than one SAP associated with it for a specific layer, just as a station may
have more than one session active via one SAP. Any link level connection of
one station to another is known as a LINK, while at each end of a link,
associated control support is referred to as a LINK STATION. In the remainder
of this section, a LLC service access point will be referred to as a SAP.
For example, a LAN station may have a 802.2 LLC level session through SAP
X'04' with SAP X'08' in a token-ring attached 3174-01L, and concurrently be in
session with a file server in another LAN station through SAP X'F0'. Logical
link control provides the capability to manage these independent sessions.
+------------------------------------------------------------------------------
]
] Station A Station B Station C
] x y
] ] ]
] V V
] ] ] ] ] ] ]
] ] (A1) ] ](B1) (B2) (B3)] ] (C1) (C2) ]
] ]---------( )-] ]( )-( )-( )] ]-( )-( )----]
] ] LLC o ] ] o A ] ] A LLC ]
] ]-----------o--] ]--o----+------] ]--+-----------]
] ] MAC o ] ] o ] ] ] ] MAC ]
] ]-----------o--] ]--o----+------] ]--+-----------]
] ===] Phys. o ]====] o ] ]=======] ] Phys.]====
] *-----------o--* *--o----+------* *--+-----------*
] oooooooooooo *-----------------*
] Medium
] ==================================================================
+------------------------------------------------------------------------------
Figure 39. SAPs and LLC Connections
In Figure 39 under heading "3.2.2 Logical Link Control Sublayer", application
x is using SAP (B1) to communicate with a resource in station A. Assume that
this resource is a printer. A connection is set up between SAP (B1) and (A1).
Once this connection is established, output can be sent to the printer. At
the same time, application y, using SAP (B2), can have a connection with a
resource in station C using SAP (C1). The use of Service Access Points allows
the applications and upper layer protocols to concurrently access the
adapter(s) and media of the LAN.
Another way to describe an LLC SAP is to define it as an architected code
point between DLC and the upper layers identifying an application to the LLC
sublayer. The IEEE standard defines a number of SAP addresses, while the IBM
Token-Ring Network Architecture uses a number of user-reserved SAP addresses
to interface with IBM proprietary protocols. Before detailing the architected
SAP values in use on IEEE LANs, it is necessary to describe bit ordering
techniques. These have equal applicability to MAC addresses, in fact, one of
the easiest places to see the various techniques is to trace TCP/IP Address
Resolution protocols.
3.2.2.1.1 Canonical Bit Ordering: In the documentation produced by the
standards bodies, a different method of bit ordering is used to that commonly
accepted in the IBM world. The "standard" way of describing the order of bits
in a byte is known as CANONICAL BIT ORDERING and is the reverse of the way IBM
usually writes binary values. Depending on whose documentation is being
studied, and when trying to relate the sets of documentation to each other,
confusion may result. Also, when tracing a token-ring, a field containing a
station address field may be present within the data. This field may be in
canonical bit order, and has to be translated into the IBM order if sense is
to be made of it. Hence an explanation of this technique here. For example:
The decimal number 71 has a hexadecimal value of 47. IBM would write this
value, in binary, as
0100 0111
In canonical form, this same binary value would be read right to left, so as
to look like:
1110 0010
If we take a much longer bit stream, such as a station MAC address, then:
1 0 0 0 5 A 1 7 5 8 6 9
0001 0000 0000 0000 0101 1010 0001 0111 0101 1000 0110 1001
Canonically
0000 1000 0000 0000 0101 1010 1110 1000 0001 1010 1001 0110
0 8 0 0 5 A E 8 1 A 9 6
The order of the bytes remains the same, but each byte has its binary digits
read from right to left when converting from the IBM form to the canonical
form.
Canonical bit ordering has particular relevance when looking at SAP values,
and the reasons why SAP values are numbered in the way in which they are.
3.2.2.1.2 SAP Values The structure of the format of the source SAP and
destination SAP fields within the LLC Protocol Data Unit is shown in Figure 40
under heading "3.2.2.1.2 SAP Values".
-------------------------------------------------------------------------------
IBM bit ordering
*-----------------*
]D]D]D]D]D]D]U]I/G] Destination SAP field (DSAP)
*-----------------*
*-----------------*
]S]S]S]S]S]S]U]C/R] Source SAP field (SSAP)
*-----------------*
Canonical bit ordering
*-----------------*
]I/G]U]D]D]D]D]D]D] DSAP
*-----------------*
*-----------------*
]C/R]U]D]D]D]D]D]D] SSAP
*-----------------*
C/R Command/Response bit - 0=Command 1=Response
U User defined SAP - 0=User 1=Standard
I/G Individual/Group SAP - 0=Individual 1=Group
-------------------------------------------------------------------------------
Figure 40. SAP Fields - Bit Ordering
Since the C/R bit of the SSAP is not used as a bit within a SAP address, there
are in fact only 128 possible SSAP values. A group SAP can only be used as a
destination SAP, and is a special case of one of the 128 possible SAP values.
These 128 SAPs are then divided into two groups, each with 64 values. If the
bits of the SAP are shown in canonical order, binary values '100 0000' to '111
1111' are reserved for IEEE use. SAPs '000 0000' to '011 1111' are for users.
The pattern of SAP values can now be more easily seen, than if they were
represented in IBM bit order. Whether the Group SAP bit is on or not, makes
no difference to the allocation. A list of architected SAP addresses is
provided in Figure 41 under heading "3.2.2.1.2 SAP Values" . This shows the
SAP values in canonical form, together with their IBM hexadecimal
representation.
*---------------------------------------------------------------*
]I/G U ]IBM HEX] Definition ]
]---------------------------------------------------------------]
]IEEE SAPs U bit = 1 ]
]---------------------------------------------------------------]
] 0 0 00 0000 ] X'00' ] Null SAP ]
] ] ] ]
] 0 1 00 0000 ] X'02' ] LLC Sublayer Management ]
] ] ] ]
] 0 1 00 bbbb ] X'x2' ] Network Mgt Function ]
] ] ] ]
] 1 1 00 0000 ] X'03' ] Group LLC Sublayer management ]
] ] ] ]
] 0 1 10 0000 ] X'06' ] D.O.D. Internet ]
] ] ] ]
] 0 1 10 bbbb ] X'x6' ] National Stds Bodies ]
] ] ] ]
] 0 1 11 0000 ] X'0E' ] Proway network management - maintenance]
] ] ] and initialization. ]
] 0 1 11 0010 ] X'4E' ] Manufacturing message service (MMS) ]
] ] ] ]
] 0 1 11 1110 ] X'7E' ] ISO 8208 (X.25 PLP) ]
] ] ] ]
] 0 1 11 0001 ] X'8E' ] Proway active station list maintenance ]
] ] ] ]
] 0 1 11 1111 ] X'FE' ] OSI Network Layer protocols ]
] ] ] ]
] 0 1 01 0101 ] X'AA' ] Subnetwork Access Protocol (SNAP) ]
] ] ] ]
] 0 1 00 0010 ] X'42' ] Bridge Spanning Tree protocol ]
] ] ] ]
] 1 1 11 1111 ] X'FF' ] Global SAP ]
]---------------------------------------------------------------]
]bbbb can be anything except B'0000' ]
]---------------------------------------------------------------]
]---------------------------------------------------------------]
]IBM defined SAP values U bit = 0 ]
]---------------------------------------------------------------]
] 0 0 10 0000 ] X'04' ] SNA Path Control Individual ]
] ] ] ]
] 1 0 10 0000 ] X'05' ] SNA Path Control Group ]
] ] ] ]
] 0 0 00 1111 ] X'F0' ] NETBIOS ]
] ] ] ]
] 0 0 10 1111 ] X'F4' ] LAN Management Individual ]
] ] ] ]
] 1 0 10 1111 ] X'F5' ] LAN Management Group ]
] ] ] ]
] 0 0 01 1111 ] X'F8' ] IMPL ]
] ] ] ]
] 0 0 11 1111 ] X'FC' ] Discovery ]
] ] ] ]
] 0 0 11 1011 ] X'DC' ] Dynamic Address Resolution ]
] ] ] (Name Management) ]
] 0 0 10 1011 ] X'D4' ] Resource Management ]
*---------------------------------------------------------------*
Figure 41. Architected LLC Service Access Points
Figure 42 under heading "3.2.2.1.2 SAP Values" illustrates how SAPs might be
implemented in a station to provide logical interface points for different
higher-layer protocols (including SNA path control), to acquire data link
control services.
+------------------------------------------------------------------------------
]
]
]
] ] SNA System ] OSI ]
] ] ] System ]
] ]- - - - - - - - * ]- - - - ]
] ] SNA DLC Manager] ] ]
] LLC ] ] ISO ]
] User ] LLC Sub-layer ] ] Network]
] ] Management (DLC Manager / Path Control) ] Layer ]
] ] ] ] ]
] = = = = = = *--* = = = = = = =*--*= = =*--*= = =*--*= = = =*--*= =
] ] ] ]02] ] ] ]]04] ]08] ]0C]] ] ]FE] ]
] LLC ] ] *--* ] ]*--* *--* *--*] ] *--* ]
] Sub- ] *------* ] *----------------------* ] ]
] layer ]group SAP 03 group SAP 05 ] ]
] ] ] ] ]
] *--------------------------------------------------------*
]
]
]
+------------------------------------------------------------------------------
Figure 42. LLC Service Access Points and SNA
3.2.2.2 IEEE 802.2, ISO 8802-2
To satisfy the broad range of potential applications, IEEE 802.2 logical link
control defines different types of operation. These types of operation, each
representing a number of link-level services, are combined in several SERVICE
CLASSES.
o TYPE 1, CONNECTIONLESS OPERATION
The Connectionless services are also referred to as USER DATAGRAM mode of
operation. In this mode, there is no data link connection establishment
between the SAPs of the end stations before actual information frame
transmission starts. At the LLC level there is no guaranteed delivery of
frames to the destination station. There is no correlation between frames
in a particular data transmission, and the sender does not expect an
LLC-level acknowledgement. In this mode of operation, higher layer
services must provide recovery and frame sequencing.
o TYPE 2, CONNECTION-ORIENTED OPERATION
Prior to data exchange, a data link connection must be established, for
which a LLC Type 2 control block has been defined. These control blocks
together with the associated delivery and error recovery services are
called LINK STATIONS.
The line protocol which is maintained on an established link between end
stations is HDLC ASYNCHRONOUS BALANCED MODE EXTENDED. In this way
reliable end-to-end service for high traffic rates is provided.
Essentially, the LLC Type 2 services may be summarized as:
- Sequence numbering of data frames at the data link layer.
- Error detection and basic recovery.
- Flow control.
This also includes an LLC-level acknowledgement, for which a WINDOW SIZE
of up to 127 outstanding frames may be sent before expecting
acknowledgement from the destination station.
o TYPE 3, ACKNOWLEDGED CONNECTIONLESS OPERATION
This third mode of operation has been proposed as an enhancement to the
international standard.
In this mode, there is no connection establishment prior to data exchange,
as in datagram operation. But link-level acknowledgement, consistent with
the window size in use, is expected by the sending station.
Type 3 operation seems particularly appropriate for LANs which experience
traffic patterns with unpredictable bursts in message rates, such as a LAN
with file servers or backbone traffic.
Currently two classes of LLC operation are defined within the IEEE 802.2
Standard based upon the above Type 1, Type 2, and Type 3 services:
o Class I - has Type 1 service only
o Class II - has Type 1 and Type 2 service.
o LLC Service
However, as more types of LLC service are defined, there is a move to drop the
concept of classes, due to the greater number of combinations of service, and
hence classes, that would be created.
3.2.2.3 LLC Protocol Data Unit
Figure 43 under heading "3.2.2.3 LLC Protocol Data Unit" shows the format of
a logical link control protocol data unit (LPDU) and the code bits for
identifying each of the LLC services for both connectionless and
connection-oriented modes of operation.
+------------------------------------------------------------------------------
]
]
] ]<---- LLC protocol data unit ---->]
] *------------------------------------------------------------------*
] ] MAC Header ] RI ]DSAP]SSAP] CF ] Information Field ] MAC Trailer ]
] *------------------------------------------------------------------*
] ] ] ] ] ]
] Routing --* ] ] ] *-- 0, 1, or more bytes
] Information ] ] *----- Control Field, 1 or 2 bytes
] (optional) ] ] bit 0 7 0 7
] ] ] *---------------------------------*
] Destination SAP ----* ] I-Format ] ] ] N(S)] ] ]0] ] ] N(R)] ] ]P/F]
] *-----------------* ] *---------------------------------*
] ]D]D]D]D]D]D]U]I/G] ] *---------------------------------*
] *-----------------* ] S-Format ]0]0]0]0]F]F]0]1] ] ] N(R)] ] ]P/F]
] 0 1 2 3 4 5 6 7 ] *---------------------------------*
] *-----------------* ] *-----------------*
] ]S]S]S]S]S]S]U]C/R] ] U-Format ]M]M]M]P/F]M]M]1]1]
] *-----------------* ] *-----------------*
] Source SAP --------------* bit 0 7
]
]
] *--------------------------------------------------------------------*
] ] Connection-Oriented Services ]
] ]--------------------------------------------------------------------]
] ] Supervisory Function bits ]0 0]RR ]Receive Ready (cmd/resp) ]
] ] (S - Format) ]0 1]RNR]Receive Not Ready (cmd/resp) ]
] ] ]1 0]REJ]Reject (cmd/resp) ]
] ]--------------------------------------------------------------------]
] ] Modifier Function bits ]0 0 0 1 1]DM ]Disconnect Mode resp ]
] ] (U - Format) ]0 1 0 0 0]DISC ]DISConnect cmd ]
] ] ]0 1 1 0 0]UA ]Unnumbered Ackn resp ]
] ] ]0 1 1 1 1]SABME]Set Asynch Balanced Mode ]
] ] ] ] ] Extended cmd ]
] ] ]1 0 0 0 1]FRMR ]FRaMe Reject resp ]
] ] ] ] ] (5 byte I-field) ]
] ] ]1 0 1 1 1]XID ]eXchange IDentification ]
] ] ] ] ] cmd/resp (3 byte I-f)]
] ] ]1 1 1 0 0]TEST ]TEST cmd/resp ]
] ] ] ] ] (optional I-field) ]
] ]--------------------------------------------------------------------]
] ] Connectionless Services ]
] ]--------------------------------------------------------------------]
] ] Modifier Function bits ]0 0 0 0 0 0]UI ]Unnumbered Inform. cmd ]
] ] (U - Format) ]1 0 1 1 1]XID ]eXchange IDentification ]
] ] ] ] ] cmd/resp (3 byte I-f)]
] ] ]1 1 1 0 0]TEST ]TEST cmd/resp ]
] ] ] ] ] (optional I-field) ]
] *--------------------------------------------------------------------*
]
]
+------------------------------------------------------------------------------
Figure 43. IEEE 802.2 LLC Protocol Data Unit Format
Each LPDU consists of a destination and source SAP address field, a control
field and an information field of zero or more bytes. The following table
lists the abbreviations used in this figure:
o D = Destination SAP bits.
o S = Source SAP bits.
o U = User-defined address bit (B'1' if IEEE defined).
o I/G = Individual (B'0') or group (B'1') SAP address.
o C/R = Command (B'0') or response (B'1') LPDU.
o N(S) = Transmitter send sequence number.
o N(R) = Transmitter receive sequence number.
o P/F = Poll/final bit.
o F = Supervisory function bit.
o M = Modifier function bit.
3.2.2.3.1.1 DSAP Field: The destination service access point (DSAP) address
field identifies one or more service access points for which the information
is intended.
This eight-bit field contains one bit to identify whether the address is an
individual or a group address, and seven address bits. Two group addresses
have specific IEEE standard definitions:
o Global address.
If the DSAP field contains B'11111111', it is a global address (the frame
is addressed to all SAPs).
o Null address.
If the DSAP field contains B'00000000', it is a null SAP address,
applicable only to connectionless operation. It supports response to a
TEST or XID command in those situations where no SAP has yet been
activated in the remote node.
3.2.2.3.1.2 SSAP Field: The source service access point (SSAP) address field
identifies the service access point which sent the frame.
This eight-bit field contains a bit to identify whether the frame is a command
or a response, and seven address bits. The command/response bit is not
considered as part of the source SAP address in a received frame.
The low-order bit in the seven address bits of both DSAP and SSAP indicate an
IEEE defined SAP address when set to B'1'.
3.2.2.3.1.3 Control Field: The control field is a one- or two-byte field
used to designate command and response functions. It contains sequence
numbers when required. The control field is very similar to the HDLC control
field. Its contents for different LLC services are outlined in Figure 43
under heading "3.2.2.3 LLC Protocol Data Unit".
3.2.2.3.1.4 Information Field: The information field contains zero or more
bytes of information, depending on the particular LLC service contained in the
control field.
A LLC PDU is invalid under the following conditions:
o It is identified by the physical layer or MAC sublayer as being invalid.
o The frame is not an integral number of bytes in length.
o It does not contain properly formatted address fields.
o The frame contains mandatory fields that are out of order.
o The frame length is less than three bytes for a one-byte control field, or
less than four bytes for a two-byte control field.
Invalid protocol data units are ignored.
3.3 LAN Standards Summary
IBM participates in most of the major LAN standards making organizations,
including IEEE, ISO and ANSI, at international and local level. IBM has
provided many submissions in the area of LAN standardization, and has taken
steps to ensure that IBM LAN products support or closely follow these
standards. IBMs active role in standards activity is particularly
demonstrated by the SRT Bridge suggestions to IEEE 802.1D.
IBM participates in the Manufacturing Automation Protocol (MAP) project, and
has been active in the development of the FDDI standard, particularly within
the area of the complex protocols of station management.
Although the IBM PC Network adopts some of the concepts that are used by IEEE
802.3, it cannot be considered a standard product, differing from IEEE 802.3
in terms of the media and transmission speeds utilized. However, the IBM
Token-Ring Network fully conforms to (and enhances) the IEEE 802.5 and 802.2
standards. IBM has participated in and supported the following enhancements
to the IEEE 802.5 specifications:
o 16 Mbps as a standardized transmission speed
o Early token release as an option
o Source routing bridge protocols.
These proposals are in the process of being approved for inclusion in the
international standard.
IBM devices attach to IEEE 802.3, IEEE 802.4 and IEEE 802.5 networks, and
implement IEEE 802.2 protocols. IBM has said that it will, in time, support
ISO 9314 FDDI networks.
IBM is supporting ISO management on its LANs, with the IBM LAN Station Manager
product, and using the OSI CMIP protocol to convey management information. IBM
is supporting international standards, and the advent of the concept of
HETEROGENEOUS LAN MANAGEMENT, a joint project between IBM and 3Com, means that
a managed LAN environment, no matter on what topology or technology it is
based, could become a practical possibility.
4.0 Structured Wiring
Standard wiring between telephone plugs in an office area and (automated)
patch panels has been a common practice for some time. This has not generally
been true for data communications where workstations have been connected from
their particular locations to control units in other locations by pulling
dedicated coaxial or twinaxial cables. The choice of cable type for data
communications has also been rather restrictive with respect to the variety of
devices that may be attached to a particular type of cable. In an
unstructured wiring environment, every relocation or change in type of a
device may require considerable planning and installation effort due to these
factors.
The emergence of intelligent workstations has led to significant changes in
application connectivity: from the traditional hierarchical host-workstation
connectivity to a mixture of hierarchical and peer-to-peer connectivity such
as that between intelligent workstations used as LAN requesters and servers.
For these reasons, a structured and consistent wiring approach to address the
cabling requirements for a large number of different devices and their
attachments has become important to many organizations.
4.1.1 Star-Wired Environments
The basic idea of structured wiring is to pre-wire work areas with
high-quality cables running to central locations, called WIRING CLOSETS where
every permanent wire ends at a DISTRIBUTION PANEL. The overall purpose is to
simplify moves, changes and reconfigurations.
Any change in the location or characteristic of a workstation would usually
require only a different connection via jumpers at the distribution panel.
More substantial changes in the topology of the attachments would usually be
accomplished by adding components in the wiring closet to support the new
wiring or attachment requirements, or by pulling additional wire (such as
fiber cable) through the shorter and more accessible runs between the wiring
closets. Thus the use of a star-wired approach minimizes the costs of
maintenance by ensuring that the areas subject to the greatest change are the
areas that are the most accessible and least expensive to change from an
equipment and labor standpoint. The same structuring also provides simpler
problem determination and recovery procedures, since the connection points are
clustered in the wiring closets where they are more readily identified and
where backup ports can be accessed.
4.1.2 Other Wiring Environments
Other requirements, such as very small networks or the need for broadband
transmission (to support both CATV and data transmission) may lead to
different wiring approaches from the star-wired approach identified above.
IBM PC Network implementations provide cabling kits or accessories to address
either of these requirements.
The IBM PC Network (Broadband) LAN is a 2 Mbps CSMA/CD driven local area
network, offered with IBM-supplied cabling kits, referred to as SHORT, MEDIUM
AND LONG DISTANCE KITS. The cable type used is CATV cable. The wiring is
dedicated for use in the PC Network (Broadband) and is not used for other
devices or transmission. With custom wiring using more sophisticated
transmitters, the other CATV cable channels may be used for other purposes
such as closed-circuit TV. For more details, see "5.1.3 IBM PC Network
(Broadband) Components."
IBM PC Network Baseband is a low-cost 2 Mbps CSMA/CD driven local area
network, using twisted pair wire to connect its LAN stations into one or more
daisy-chains. A PC Network Baseband adapter cable may be obtained to take
advantage of the IBM Cabling System in wiring this LAN. "5.2 IBM PC Network
Baseband" provides more details on this LAN offering.
4.2 IBM Cabling System
The IBM Cabling System provides specifications for components and their use in
structured pre-wiring of buildings. The components include cables,
connectors, wall faceplates, and wiring closet distribution panels.
While IBM helped design specifications for the cabling system, manufacturing,
distribution, installation, and maintenance is done by third parties.
The IBM Cabling System provides the elements of a common wiring system capable
of supporting most existing and future IBM workstation equipment, in addition
to several host processors. Today, the IBM Cabling System consolidates coax,
twinax and telephone twisted pair wire transmission support into a
multi-purpose quality twisted pair wire (data grade shielded twisted pair).
The wiring specifications also permit cost-effective use of fiber cables for
backbone wiring. In addition to providing bandwidth for future traffic
requirements, the use of fiber backbone cables enables an establishment to
install wiring with the potential to support possible FDDI LAN requirements.
The IBM Cabling System components are designed to support a concept of
structured wiring minimizing the impact of replacement or relocation of
equipment. It may be used to pre-wire buildings for all equipment, or to
provide additional cable runs for new equipment in existing buildings. While
the IBM Cabling System provides specifications for several different types of
cable (some of which also support voice transmission), other types may also be
supported by the cabling system components subject to attenuation and
signalling characteristics. Figure 44 under heading "4.2 IBM Cabling System"
illustrates the eight cable types defined within the IBM Cabling System and
described in IBM cable planning reference manuals.(10 under heading "4.2 IBM
Cabling System")
---Footnote---
(10)
USING THE IBM CABLING SYSTEM WITH COMMUNICATION PRODUCTS (GA27-3620).
IBM CABLING SYSTEM PLANNING AND INSTALLATION GUIDE (GA27-3361).
--------------
+------------------------------------------------------------------------------
]
] *---------------------------------------------------------------*
] ] ] ] ] ] ] ]
] ] Cable ]Conductor] Pairs ] Shielding ]Outdoor ] Use ]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 1 ]Solid ]Two AWG 22] Braided or] Yes ] Main Paths ]
] ] ]Copper ]Twisted ] Corrugated] (1) ]Work Area to ]
] ] ] ] ] ] ]Wiring Closet]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 2 ]Solid ]Two AWG 22] Braided ] No ]Work Area to ]
] ] ]Copper ]Twisted +] ] ]Wiring Closet]
] ] ] ]Four TTP ] No ] ] ]
] ] ] ] (AWG 22) ] ] ] ]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 3 ]Solid ]Four TTP ] No ] No ]Work Area to ]
] ]Specs ]Copper ] (AWG 26) ] ] ]Wiring Closet]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 5 ]Fiber ]Two Fiber ] No ] Yes ] Main Paths ]
] ] ]Optics ] 100/140 ] ] ](inter Bldg) ]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 5J]Fiber ]Two Fiber ] No ] Yes ] Main Paths ]
] ] ]Optics ] 50/125 ] ] ](up to 500m) ]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 6 ]Stranded ]Two AWG 26] Braided ] No ]Work Area and]
] ] ]Copper ]Twisted ] ] ]Wiring Closet]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 8 ]Solid ]Two AWG 26] Braided ] No ]Work Area ]
] ] ]Copper ]Parallel ](each pair)] No ]Under carpet ]
] ]-------+---------+----------+-----------+--------+-------------]
] ]Type 9 ]Solid ]Two AWG 26] Braided ] No ] Main Paths ]
] ] ]Copper ] ] ] ] (plenum) ]
] *---------------------------------------------------------------*
]
] Note 1. Plain, Plenum and Outdoor versions
] TTP = Telephone Twisted Pair
] AWG = American Wire Gauge
]
]o Type 1 - Shield around two twisted pair of No. 22 AWG solid copper
] wire.
]o Type 2 - Same as the Type 1 cable with the addition of four twisted
] pair No. 22 AWG telephone conductors.
]o Type 3 - telephone twisted pair specification, No. 22 or No. 24 AWG
] solid copper wire.
]o Type 5/5J - Two optical fiber conductors, 100/140 and 50/125 (fiber
] core/cladding size in microns), other sizes may also be used (for
] example, 62.5/125 micron).
]o Type 6 - Shield around two twisted pair of No. 26 AWG solid copper
] wire used for patch cables.
]o Type 8 - Undercarpet cable, two pair No. 26 AWG solid copper wire,
] with shielding.
]o Type 9 - Two No. 26 AWG solid copper wire, offered as a lower cost
] alternative to type 1 plenum (11 under heading "4.2 IBM Cabling System")
+------------------------------------------------------------------------------
Figure 44. Cable Types
---Footnote---
(11) Plenum refers to space in the ceiling of a building floor frequently used
to run cable ducts. In such cases, local regulations may require the
installation of plenum cable that has a teflon coating instead of regular PVC.
--------------
There are many advantages to structured wiring systems incorporating
general-purpose cable and components:
o Minimal additional cost of adding a new terminal. The provision of cable
and wall outlets at each workplace makes it possible to "plug in" the new
equipment.
o Easier and less disruptive movement of equipment. With a wiring closet
structure, most changes are made in the wiring closet, minimizing the need
to pull or relay new cable. Where changes are required, they can usually
be confined to the more accessible wiring closets and shorter cable runs
between the wiring closets.
o Simpler problem determination and maintenance. The knowledge of which
cable is installed and where helps to minimize cost and time spent
associated with maintenance or problem determination.
A key element in problem determination at this level is proper documentation
of the cabling system installation, see USING THE IBM CABLING SYSTEM WITH
COMMUNICATION PRODUCTS. Also refer to recommendations for labelling schemes,
worksheets and charts to document the installation in the IBM CABLING SYSTEM
PLANNING AND INSTALLATION GUIDE.
The fiber cabling accessories, attachments, and planning guidelines supplied
by IBM support the most popular multimode optical fiber specifications,
including 50/125 micron, 62.5/125 micron, 85/125 micron, and 100/140 micron
cables. Because the FDDI LAN specifies 62.5/125 micron optical fiber, many
customers may wish to use 62.5/125 micron multimode optical fiber cable in
their token-ring network installations. Alternatively, they could use the
50/125 micron specification for those countries where 62.5/125 micron is not
commonly used or available. 85/125 or 100/140 micron optical fiber cable are
supported, but are not recommended for new installations by customers
considering use of FDDI for future applications.
For IBM recommendations on fiber optic cabling, please refer to "2.1.1.2 Fiber
Optic Cable."
5.0 IBM Local Area Network Solutions
IBM provides three local area network implementations:
o IBM Token-Ring Network
o IBM PC Network (Broadband)
o IBM PC Network Baseband.
IBM has also announced products that implement station attachment to LANs
based upon the Manufacturing Automation Protocols (MAP) 3.0 specification,
including IEEE 802.4 networks.
IBM also provides attachment of some of its products to Ethernet Version 2 and
IEEE 802.3 networks. Products that may attach include PC and PS/2 workstations
running DOS or OS/2 operating software, PC RTs and RS/6000 processors, and the
IBM 8209 LAN bridge that may be used to link Ethernet/IEEE 802.3 LAN segments
to token-rings. Gateway support to IBM hosts is provided by both the IBM 8232
and IBM 3172 Interconnect Controllers.
5.1 IBM Token-Ring Network
The IBM Token-Ring Network uses baseband transmission at either 4 Mbps or 16
Mbps implementing the IEEE 802.5 standard MAC protocols and the IEEE 802.2
Class II standard LLC protocols. At 16 Mbps, the IBM Token-Ring Network
implements the early token release option.
Devices operating at 4 Mbps and others operating at 16 Mbps cannot be
intermixed on the same LAN segment. Any attempt to insert a device into the
ring at the wrong data rate will cause the ring to fail for several seconds
due to the recovery process.
The IBM Token-Ring Network uses a star-wired logical ring topology. Devices
on the network are physically attached to wiring concentrators. The wiring
concentrator or access unit may be passive, such as the IBM 8228, known as a
MULTISTATION ACCESS UNIT (MSAU). or active devices such as the IBM 8230
Controlled Access Unit (CAU). The access unit is used as a point to control
physical access to the ring, and to provide a method to bypass faulty or
inactive devices. Use of an access unit avoids the following problems
inherent in a physical ring topology:
o RELIABILITY
Failure of a ring station may disrupt the ring due to the loss of signal
regeneration. Without a station bypass mechanism, the failure of a single
station could stop all ring traffic, affecting all other ring devices. A
cable fault between adjacent LAN stations in a physical ring topology can
also stop normal operation of the LAN.
o SERVICEABILITY
Changing the configuration of the network (or even adding a new LAN
station) could be disruptive if a controlled insertion process is not
implemented.
Using a wiring concentrator can alleviate or remove these problems. For
example, having a wiring concentrator that bypasses any inactive station means
that when a station fails or a cabling fault occurs in an attaching cable, the
ring is not disrupted. Stations can be added or removed from the ring while
it is running without affecting other users in the network.
5.1.1 IBM Token-Ring Network Components
5.1.1.1 Cabling Components
The IBM Token-Ring Network is designed to make use of a structured wiring
system, like the IBM Cabling System mainly because of its star-wired ring
topology. Maximum flexibility may be obtained when a building is wired with
conveniently located wiring closets containing wiring patch panels and
token-ring access units.
The IBM Token-Ring Local Area Network can be used with all the cable types
referred to in Figure 44 under heading "4.2 IBM Cabling System" when
installed in conformance with the planning and installation specifications of
the IBM Cabling System as described in the IBM TOKEN-RING NETWORK INTRODUCTION
AND PLANNING GUIDE. Type 1 or 2 cables should be used for lobes (the
connection from the wall outlet to the workstation). Cable Types 1, 2 and 5
can be used for the links between the access units.
In an IBM Token-Ring Network operating at 4 Mbps, Type 3 Telephone Twisted
Pair (TTP) cable can be used to connect device attachment locations to access
units in a wiring closet provided the Type 3 Cable Specifications for distance
and number of station attachments are met, and the cable is not subject to
electrical noise. Type 3 TTP wire can be mixed with data grade media twisted
pair wire within a single physical ring but cannot be used to wire a 16 Mbps
token-ring LAN segment.
Figure 45 under heading "5.1.1.1 Cabling Components" provides a schematic
diagram of IBM Token-Ring Network stations connected via IBM Cabling System
wiring and attachments. The first part shows the use of data grade media
twisted pair; the second illustrates installation requirements when using
voice grade (telephone) twisted pair wire.
+------------------------------------------------------------------------------
]
] ** Lobe :
] ]] Type 1/2/9 inter-Wiring
] ]] *-----------* Closet : Type 1/9/5
] ]] ] *---]----------------<-------
] ]] ] ] ] from Previous
] ]] ] ] ] *------>-------
] ]] ] *]---]---------]* to Next
] *----* ]] ] *-. .*. . . . . .-<-*
] ] ] ]] ] ]]. .]. . . . . .] ]
] ] ] ]] ] ]]. .]. . . . . .] ]
] *------* ]] ] V]. .]. . . . . .] ]
] ] ]-----* Device ]] ] ]]. .]. . . . . .] ]
] *------* ] Adapter]] ] ]]---]-----------] ] Type 6
] ] Cable ]] ] ]]---]-----------] ] (patch
] *--------]]---* AU *+. .*. . . . . .+* ] cable)
] ]] ]---------------]] ]
] ]] ]---------------]] ]
] ]] AU *+. . . . . . . .+]-*
] ]] ]]---------------]]
] ]] *--------<--------*
] ]] ] ]
] ]] ] ]
] -------- - -
]
] DATA GRADE MEDIA TWISTED PAIR, 4 MBPS OR 16 MBPS
] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
]
]
] Type 3 Voice **
] Jumper Cables Pairs ]]
] ] Telephone ] ]] Type 3 specs TTP
] *-----------------* Jumper ] ]] ]
] ] *------* ] Wire-* ] ]] ] 30 Mhz
] *----]-------]--* ] ] *--* ] *--* ]] ] Media
] ]. . .*. . . ] .] ] *----] ]------] ]---]]--------* Filter
] ]. . .]. . . .*.] ] ] ]------] ] ]] ] /
] ]. . .]. . . .].] ] ] ]------] ] ]] *-* *----*
] ]. . .]. . . .].] *-------] ]------] ]---]]---* *-* ] ]
] ]. . .]. . . .].] *--* *--* ]] ] ] ] ]
] ]-----]-------]-] Type 66 ]] ] *---------*
] ]-----]-------]-] Termination ]] ] ] ]
] ]. . .*. . . .*.] AU Blocks ]] *-* *------*
] ]---------------] ]] *-*
] ]---------------] ]] ] *----*
] ]o o o o o o o o] ]] ] ] ]
] ]---------------] ]] ] ] ]
] ] ] ]] *------------*
] ] ] ]] ] ]
] - - -------- *------*
]
] VOICE GRADE TELEPHONE TWISTED PAIR, 4 MBPS ONLY
]
+------------------------------------------------------------------------------
Figure 45. IBM Cabling System in the IBM Token-Ring Network
o A maximum of 33 IBM 8228 Multistation Access Units is supported on a
single ring running at 4 Mbps or 16 Mbps. The maximum number of
concurrently active devices on a 4 or 16 Mbps Token-Ring Network is 260.
o When Telephone Twisted Pair wire is being used a media filter must be
installed between the station adapter card and the telephone jack in the
wall. Also the maximum number of concurrently active devices is 72 (TTP
is supported by IBM running at 4 Mbps only).
Two type 66 termination blocks are recommended to provide isolation
between the data and voice pairs of the telephone wire. Between a data
termination block and the distribution panel a Type 3 jumper cable is
used.
The Telephone Twisted Pair wiring option may not be available in all
countries due to PTT restrictions.
o For further information, refer to the following documentation:
- IBM CABLING SYSTEM PLANNING AND INSTALLATION GUIDE
- IBM TOKEN-RING NETWORK INSTALLATION GUIDE
- IBM TOKEN-RING NETWORK INTRODUCTION AND PLANNING GUIDE
- IBM TOKEN-RING NETWORK TELEPHONE TWISTED-PAIR MEDIA GUIDE
- IBM TOKEN-RING NETWORK TELEPHONE TWISTED-PAIR MEDIA GUIDE, European
version.
In addition to the specifications for numbers of attachments identified above,
rules are also provided for maximum distances between stations which drive
signals, and wiring closets (that is, maximum lobe distances) and between
wiring closets.
The number of stations attached to a token-ring LAN may be increased
indefinitely by using token-ring bridges. For instance 4 Mbps ring segments
may be interconnected with 16 Mbps segments via token-ring bridges; or voice
grade (telephone) twisted pair wired ring segments may be interconnected with
ring segments for which data grade media twisted pair wire has been used.
IBM Token-Ring Network bridges are discussed in "6.0 LAN Segments
Interconnection."
5.1.1.1.1 IBM 8228 Multistation Access Unit The IBM 8228 Multistation Access
Unit supports up to eight LAN stations via device attachment ports and the
interconnection of multiple 8228 access units via ring-in, ring-out ports.
The 8228 is a passive wiring concentrator that can bypass attached devices by
reacting to the presence (or absence) of a signal from the device attachments.
Each of the eight device attachment ports includes relay mechanism which is
powered by a phantom current (12 under heading "5.1.1.1.1 IBM 8228
Multistation Access Unit") from the attaching adapter.
---Footnote---
(12) A 5 Volts DC voltage is sent by the attaching adapter over the lobe wire t
power the relay. It is called a phantom current because it is transparent to th
actual AC data signal.
--------------
In "2.2.2 Basic Token-Passing Ring Concepts," a sample wiring scheme involving
four multistation access units (referred to as passive wiring concentrators)
was described together with an illustration, in Figure 10 under heading "2.2.2
Basic Token-Passing Ring Concepts", of a few active or inserted devices as
well as several bypassed lobes.
The backup scenario shows how the backup path is used to recover the ring
traffic following a cable break between the ring-out position of one
multistation access unit and the ring-in position of the subsequent
multistation access unit. This recovery requires manual intervention to remove
each end of the cable segment from the ring-in or ring-out ports, unless a
pair of 8220 Optical Fiber Converters has been installed with the capability
to isolate the broken segment (see "5.1.1.1 Cabling Components").
5.1.1.2 IBM 8230 Controlled Access Unit
The IBM 8230 is an intelligent token-ring network wiring concentrator,
providing enhanced levels of control and reliability over passive token-ring
network wiring concentrators, such as the IBM 8228. It is designed to:
o Improve token-ring availability
o Support the IBM LAN Network Manager, see "9.3.5 IBM LAN Network Manager
V1" by:
- Supporting IBM LAN Network Manager's configuration table
- Supporting access control by reporting station insertions
- Supporting asset control in conjunction with IBM LAN Network Manager.
It is a customer set up, rack-mountable device that:
o Supports ring operation at 4 and 16Mbps
o Is able to function as a repeater in both ring directions
o Has pluggable ring-in and ring-out modules to support copper and fiber
cable
o Has token-ring MAC appearances on both the main and backup ring path
o May have its microcode loaded from the IBM LAN Network Manager, or a
diagnostic utility provided.
One IBM 8230 attaches up to 80 workstations to the token-ring local area
network (IEEE 802.5).
Each IBM 8230 consists of two types of unit:
o A Controlled Access Unit Base (Base Unit).
o And up to four Lobe Attachment Modules (LAM). Each LAM can attach 20
devices to the ring.
Although the 8230 operates at either 4 Mbps or 16 Mbps, unshielded twisted
pair wiring on the lobes is only supported at 4 Mbps. In this case, a special
media filter is required to be plugged into the control unit.
The IBM 8230 works with both the copper and optical fiber cables of the IBM
Cabling System. The ring-in and the ring-out connections are provided by two
types of pluggable module, one suitable for copper, the other for fiber
connections. Thus the ring-in (RI) and ring-out (RO) connections to an 8230
may be both copper, both optical fiber or one of each, depending on the RI and
RO options installed in the control unit.
The fiber RI and RO modules accept 50/125, 62.5/125, 85/125 and 100/140
multi-mode optical fiber cables. The optical fiber cable connections are via
standard "Mini-BNC" optical fiber connectors.
NOTE: The 8230 is shipped with two copper RI/RO modules installed as standard
equipment. The optical fiber modules must be ordered as an optional feature.
The 8230 can co-exist with the following products when installed in the same
token-ring LAN: IBM 8218, 8219, 8220, 8228.
o The 8228 may be located between 8230 units but may not be connected
directly to the 8230 lobes.
o The 8218s, 8219s, and 8220s may be used in pairs in the same ring. The
8230 CANNOT serve as one half of a repeater pair for any of these devices.
When used without associated Lobe Attachment Modules (LAM), it can operate as
a repeater, at either 4 Mbps or 16 Mbps, using optical fiber or copper media.
A remote diagnostic tool will be shipped with the 8230. It comes on a DOS
bootable diskette with each 8230. This tool is for use prior to the
availability of IBM LAN Network Manager and generally for those customers who
do not have IBM LAN Network Manager installed. The RDT will support up to 16
8230's on a local segment. It can be used for diagnostics and problem
determination purposes, and for remote program update. The top level 8230
microcode is updatable using the remote diagnostic tool or the Remote Program
Update (RPU) enable facility of IBM LAN Network Manager in combination with a
program update utility.
Figure 46 under heading "5.1.1.2 IBM 8230 Controlled Access Unit" shows the
IBM 8230 on an IBM Token-Ring Network.
-------------------------------------------------------------------------------
*------------*
] IBM 8230 ]
] ]
*--------------------] RO RI ]---------------------------*
] ] ] ]
] ] ] ]
] *------------* ]
]<----- Main Ring Path (fiber or copper) copper -->]
] ]
] *-------------------------------+--*
] ] ] ]
] ] RI IBM 8228 ]RO]
] ] ] ]
] ] *-* *-**-**-**-**-**-**-**-* *-* ]
] ] *-* *-**-**-**-**-**-**-**-* *-* ]
] *--+------------------------+------*
] Standard Patch Cable --->] ]
] *------------* ]
] ] ]
] ] ]
] *---------------------------------------* ] ]
] ] RI RO ] ] ]
] ] *---* *---* ] ] ]
*--+--] ] ] ]--+--* ]
] *---* *---* ] *-------*
] Control Unit ] ] ]
] ] ] Node ]
] ] ] ]
]------------- IBM 8230 ---------------] *-------*
] ]
] Lobe Attachment Module LAM ]
]*-* *-* *-* *-* *-* *-* *-* *-* *-* *-*]
]*-* *-* *-* *-* *-* *-* *-* *-* *-* *-*]
] ]
]*-* *-* *-* *-* *-* *-* *-* *-* *-* *-*]
]*-* *-* *-* *-* *-* *-* *-* *-* *-* *-*]
*-------------+-------------------------*
]<------- Lobe
*-------*
] ]
] Node ]
] ]
*-------*
NOTE: Standard IBM Cabling System patch cables are used to connect the
control unit and the LAMs (ICS version) to distribution panels.
-------------------------------------------------------------------------------
Figure 46. IBM 8230 on IBM Token-Ring Network
5.1.1.2.1 Hardware Description: The main parts of the control unit are power
supply and logic unit. The logic unit contains microprocessors and their
associated microcode to direct the operation of the LAMs and to interact with
the token-ring network for data transfer and communication. The power supply
provides DC power for the control unit and the associated LAMs.
The LAM provides the interface to the lobe wiring connecting the workstations
to the token-ring network. It is available in two versions. One has IBM
Cabling System (IEEE 802.5) connectors, the other RJ45 connectors. It is
possible to mix the LAM types within a single 8230.
Each outlet of the LAM has a LED for status indication:
o Off = Lobe enabled but not inserted.
o On solid = Lobe enabled and inserted.
o Blinking = Lobe disabled by 8230 or IBM LAN Network Manager.
NOTE: An ENABLED lobe is one on which a station could insert. A station
attempting to insert on a DISABLED LOBE will not get onto the ring. Lobes are
disabled and enabled by command from the IBM LAN Network Manager, or the CAU
upon discovering a fault in this area.
Attaching LAMs are "hot-pluggable", that is, they may be connected or
disconnected from the control unit while power is applied.
The 8230 uses three individual MAC addresses. Therefore it takes up three of
the total number of adapters allowed on a single ring ( = 260 for ring cabled
with Type 1 cable).
Figure 47 under heading "5.1.1.2.1 Hardware Description" shows a functional
diagram of the IBM 8230. Reference to it should be made while reading the
following text.
The primary-in (PI) adapter is the first MAC entity on the main ring. All
token-ring data comes through PI just after entering the ring-in connector
when the ring is operating normally. PI reclocks and retransmits data as any
typical station on the token-ring network. The primary-out (PO) adapter is
the second MAC entity on the main ring. All token-ring data passes through PO
just before exiting the ring-out port when the ring is operating normally. PO
reclocks and retransmits data as any typical station on the token-ring
network. The secondary (S) adapter is the MAC entity that resides on the
backup path. During normal ring operation, all data on the backup path passes
through S. S reclocks and retransmits data as any typical station on the
token-ring network.
-------------------------------------------------------------------------------
*---------------------------------------------------------------*
] *------------* *------------* ]
] ] ] *---------* ] ] ]
] ]Power Supply] ]S-Adapter] ] Controller ] ]
] ] ] *---------* ] ] ]
] *------------* ]] *------------* ]
] *--* ]
] *-] ]---------* ]
]Ring-in Wrap ] *--* ] Wrap Ring Out ]
]Module Relays ] ] Relay Module ]
] *-* *-* *-* ] ] *-* *-* ]
] ] ] ]X] ]Y] ] ] ]Z] ] ] ]
]-] ]---------]X]-]Y]----* *---]Z]-------------] ]-]
] ] ] ]X] ]Y*-----------------------]Z]-----------* ] ] ]
] ] ] ]X] ]YYYYYYYYYYYYYYYYYYYYYYYYY]Z]YYYYYYYYYYY] ] ] ]
] ] ] ]X] *-------------------------]Z]---------*Y] ] ] ]
] ] ] *--* ]X] *-* ]Z] *--* ]Y] ] ] ]
]-] ]--] ]---]X]--] ]-* *--]Z]---] ]--]Y]-] ]-]
] ] ] ] ] ]X] ] ] ] ] ]Z] ] ] *-* ] ] ]
] *-* *--* *-* *-* ] ] *-* *--* *-* ]
] ]] Filter] ] ]] ]
] *----------* ] *-* *-* *-* *-* ] *----------* ]
] ]PI-Adapter] *-] ]-] ]-] ]-] ]-* ]PO-Adapter] ]
] *----------* *-* *-* *-* *-* *----------* ]
] ] ] ] ] ]
] ] ] ] ] ]
*---------------------------------------------------------------*
Lobe attachment
Module Connectors
-------------------------------------------------------------------------------
Figure 47. IBM 8230 Functional Diagram
The function of these adapters and the wrap relays is to enable and control
automatic ring re-configuration in the event of "hard" errors on the ring
paths. How the IBM 8230 is able to react to these errors and reconfigure the
ring will be shown later in "5.1.1.2.4 Reconfiguration."
The diagram also shows the pluggable filter, required when the lobes use
unshielded cabling. The purpose of the filter is to filter the signal being
put on the first active lobe by the PI adapter.
5.1.1.2.2 Management Support: Under the control of IBM LAN Network Manager,
the 8230 provides an optional means for controlling workstation access to the
token-ring network and also for collecting realtime ring configuration or
topology data. A subset of LAN Station Manager V1 is implemented in the 8230
in order to support communication with the IBM LAN Network Manager. The
following portions of the LAN Station Manager are supported:
o Name Management. A Find/Found process will be executed to locate IBM LAN
Network Manager for the registration process. See "2.4.1.2 Registration."
o Resource Management. This means, for example, to get the operation status
of the 8230.
The 8230 supports the IBM LAN Network Manager in the areas of access and asset
control. If no LAN Station Monitor exists in the workstation, then the 8230
gives the following information directly to the IBM LAN Network Manager:
o Adapter address
o 8230 number (last 4 digits of PO MAC address)
o Lobe Attachment Module number
o Lobe number.
If there is a LAN Station Manager present in the workstation, the same
information will flow via that Station Manager to the IBM LAN Network Manager.
The 8230 transmits and receives frames from its designated MAC address which
defaults to the MAC PO address. The IBM LAN Network Manager communicates with
the 8230 using this address. In the event the IBM LAN Network Manager losing
contact, it may send frames to the RING WIRING CONCENTRATOR FUNCTIONAL ADDRESS
(X'C000 0000 1000').
5.1.1.2.3 Cabling: The 8230 supports the cable configurations in Figure 48
under heading "5.1.1.2.3 Cabling" and Figure 49 under heading "5.1.1.2.3
Cabling", provided that the cable plan meets all other requirements listed in
IBM TOKEN-RING NETWORK INTRODUCTION AND PLANNING GUIDE and IBM TOKEN-RING
NETWORK TELEPHONE TWISTED-PAIR MEDIA GUIDE.
Although a mixture of IBM Cabling System Type 1 and Type 3 lobe wiring is
supported on the same IBM 8230 Controlled Access Unit, it is recommended that
only cabling type 3 (unshielded twisted pair) be used when the LAM is fitted
with RJ45 connectors. This is due to the physical constraints of attaching
the small plugs to the thick cable. For LAMs that have the IBM data
connectors, a mixture of cabling can be used, but if there are any Type 3
lobes attached to an IBM 8230 Controlled Access Unit and its associated LAMs,
all lobes, whatever their cable type, must be fitted with the appropriate Type
3 media filter, as well as a filter installed within the 8230 Control Unit.
+----------------------------------------------------------------------------+
] Figure 48. Lobe Length ]
+-------------------------+-------------------------+------------------------+
] Data Rate (Mbps) ] Media ] Distance (m) ]
+-------------------------+-------------------------+------------------------+
] 16 ] Type 1 ] 145 ]
+-------------------------+-------------------------+------------------------+
] 4 ] Type 1 ] 375 ]
+-------------------------+-------------------------+------------------------+
] 4 ] Type 1 ] 200* ]
+-------------------------+-------------------------+------------------------+
] 4 ] Type 3 ] 100 ]
+-------------------------+-------------------------+------------------------+
] NOTE: * With any Type 3 lobes on main ring ]
+----------------------------------------------------------------------------+
Lobe distance requirements in Figure 48 under heading "5.1.1.2.3 Cabling" are
independent of the type of LAM (that is, ICS or RJ45) and number of LAMs
installed.
NOTE: Lobe length is longer than the normal recommendation because the PI
adapter reclocks and retransmits the incoming signal, thus acting as a
repeater.
+----------------------------------------------------------------------------+
] Figure 49. Ring Segment Drive Distance (Media Type 1 or Equivalent) ]
+-------------------------+-------------------------+------------------------+
] Data Rate (Mbps) ] Temperature (C) ] Distance (m) ]
+-------------------------+-------------------------+------------------------+
] 16 ] < 60 ] 200 ]
+-------------------------+-------------------------+------------------------+
] 16 ] < 80 ] 180 ]
+-------------------------+-------------------------+------------------------+
] 4 ] < 60 ] 400 ]
+-------------------------+-------------------------+------------------------+
] 4 ] < 80 ] 385 ]
+-------------------------+-------------------------+------------------------+
] 4 ] < 80 ] 300* ]
+-------------------------+-------------------------+------------------------+
] NOTE: * With any Type 3 lobes on main ring ]
+----------------------------------------------------------------------------+
The ring segment drive distances in Figure 49 under heading "5.1.1.2.3
Cabling" are the maximum distances supported between IBM 8230s when there is
no intervening equipment. Paired repeaters inserted between 8230s would
obviously increase the distances, passive equipment would reduce it. For
further information, refer to the IBM TOKEN RING INTRODUCTION AND PLANNING
GUIDE.
For definitions of Type 1 (shielded twisted pair) and Type 3 (unshielded
twisted pair) cable consult IBM CABLING SYSTEM TECHNICAL INTERFACE
SPECIFICATION.
The 8230 fiber distances are identical to the IBM 8219/8220 requirements as
specified in IBM TOKEN-RING NETWORK INTRODUCTION AND PLANNING GUIDE.
5.1.1.2.4 Reconfiguration: The IBM 8230 provides automatic ring
reconfiguration when it detects a ring fault. In that case the 8230 executes
an automatic recovery, that is, it bypasses the failing portion of the ring
segment or the failing 8230. The word fault means in this case a "hard
error". The IBM 8230 does not take action for soft errors. However, under
control of the IBM LAN Network Manager, a user can reconfigure the 8230 in
order to isolate the source of soft errors. The following new commands are
available for IBM LAN Network Manager directed recovery:
o Wrap/unwrap RI/RO
o Disable/enable lobe
o Disable/enable LAM.
The base for automatic ring configuration are the 3 MAC addresses PI,PO and S
of the 8230. PO and S are located on the ring-out (RO) side of the 8230, PI
is on the ring-in (RI) side. The 8230 can always wrap its 80 lobes to the
working portion of the ring.
The 8230 has the direct control over all ring insertion relays. This includes
the relays allowing ring stations and Lobe Attachment Modules to attach to the
ring. These relays are used to configure the broken part of the ring out of
the active ring path.
In order to follow a sample reconfiguration, reference should be made to
Figure 50 under heading "5.1.1.2.4 Reconfiguration" for a ring installation
consisting of four IBM 8230s.
-------------------------------------------------------------------------------
*----------------------* *----------------------*
] ] ] ]
] PO i i i i PI ] ] PO i i i i PI ]
] ] ] ] ] ] ] ] ] ] ] ] ] ] ] ]
*---<--+----------------------+----<---+----------------------+---<--*
] ] ] ] ] ]
] ] ] ]
] ] ] ]
] ] *----------------------* *----------------------* ] ]
] ] ] IBM 8230 "C" ] ] IBM 8230 "D" ] ] ]
] ] ] S ] ] S ] ] ]
] ] ] ] ] ] ] ] ] ]
] *-+----------------------+---X--<-+----------------------+-* ]
] ] ] ] ] ]
] RI] ]RO RI] ]RO ]
] ] ] ] ] ]
*--->--+----------------------+->-X----+----------------------+--->--*
] ] ] ] ] ] ] ] ] ] ] ] ] ]
] PI i i i PO ] ] PI i i i PO ]
] ] ] ]
*----------------------* *----------------------*
S = 8230 MAC address on backup path
PI = 8230 MAC address on primary path (RI side)
PO = 8230 MAC address on primary path (RO side)
i = ring station inserted on 8230
-------------------------------------------------------------------------------
Figure 50. Ring Cut at X
If there is an external fault between station "C" and " station "D", that is,
the cable is broken, the following will happen:
1. Ring is cut at X.
2. PI "D" beacons, showing NAUN as PO "C", with a beacon type 2 (beacon type
X'0002' - signal loss error). MAC address S "C" also beacons on the
back-up path.
3. PO "D" changes the beacon of PI "D" to a beacon type 1 (beacon type
X'0001' - recovery mode set).
4. All MAC addresses in 8230 "B" and "A" are in beacon repeat mode for both
paths.
5. "C" sees type 1 beacon from PO "D", knows his MAC address S is beaconing
and therefore wraps RO "C" using relays Y (see Figure 47 under heading
"5.1.1.2.1 Hardware Description"). The situation is now as in Figure 51
under heading "5.1.1.2.4 Reconfiguration".
-------------------------------------------------------------------------------
] ] ] ]
V A V A
] ] ] ]
] ] ] ]
] ] *----------------------* *----------------------* ] ]
] ] ] IBM 8230 "C" ] ] IBM 8230 "D" ] ] ]
] ] ] S ] ] S ] ] ]
] ] ] ] ] ] ] ] ] ]
] *-+----------------* *---+-- X --+----------------------+-* ]
] ] ] ] ] ] ] ]
] RI] ] ] ]RO RI] ]RO ]
] ] ] ] ] ] ] ]
*--->--+----------------* *---+-- X --+----------------------+--->--*
] ] ] ] ] ] ] ] ] ] ] ] ] ]
] PI i i i PO ] ] PI i i i PO ]
] ] ] ]
*----------------------* *----------------------*
-------------------------------------------------------------------------------
Figure 51. 8230 "C" wraps RO
6. Because of the wrap at RO of 8230 "C", the beacon of PO "D" will be seen
at S of 8230 "D". 8230s A and B are still present, just not shown.
7. S "D" recognizes PO "D" address in PO "D" beacon.
8. 8230 "D" wraps RI and RO, using relays X and Z (see Figure 47 under
heading "5.1.1.2.1 Hardware Description" ). The situation is now shown in
Figure 52 under heading "5.1.1.2.4 Reconfiguration".
-------------------------------------------------------------------------------
] ] ] ]
V A V A
] ] ] ]
] ] ] ]
] ] *----------------------* *----------------------* ] ]
] ] ] IBM 8230 "C" ] ] IBM 8230 "D" ] ] ]
] ] ] S ] ] S ] ] ]
] ] ] ] ] ] ] ] ] ]
] *-+----------------* *---+-- X --+----* *--------* *----+-* ]
] ] ] ] ] ] ] ] ] ] ] ]
] RI] ] ] ]RO RI] ] ] ] ] ]RO ]
] ] ] ] ] ] ] ] ] ] ] ]
*--->--+----------------* *---+-- X --+----* *--------*
*----+--->--*
] ] ] ] ] ] ] ] ] ] ] ] ] ]
] PI i i i PO ] ] PI i i i PO ]
] ] ] ]
*----------------------* *----------------------*
-------------------------------------------------------------------------------
Figure 52. Wrap RO, Wrap RI/RO
9. PO "D" sees his own beacon, knows that the ring is fixed and starts the
monitor contention process.
10. S "D" checks the internal ring, LAMs and lobes of 8230 "D". If there is
no fault within this ring, then it unwraps RO "D", by using relay Z (see
Figure 47 under heading "5.1.1.2.1 Hardware Description"). Figure 53
under heading "5.1.1.2.4 Reconfiguration" shows the recovered ring, with
all stations active.
-------------------------------------------------------------------------------
] ] ] ]
V A V A
] ] ] ]
] ] ] ]
] ] *----------------------* *----------------------* ] ]
] ] ] IBM 8230 "C" ] ] IBM 8230 "D" ] ] ]
] ] ] S ] ] S ] ] ]
] ] ] ] ] ] ] ] ] ]
] *-+----------------* *---+-- X --+----* *---------------+-* ]
] ] ] ] ] ] ] ] ] ]
] RI] ] ] ]RO RI] ] ] ]RO ]
] ] ] ] ] ] ] ] ] ]
*--->--+----------------* *---+-- X --+----* *---------------+--->--*
] ] ] ] ] ] ] ] ] ] ] ] ] ]
] PI i i i PO ] ] PI i i i PO ]
] ] ] ]
*----------------------* *----------------------*
-------------------------------------------------------------------------------
Figure 53. Wrap RO, Wrap RI. Broken cable isolated from ring path
11. S "C" and PI "D" continue monitoring the broken segment.
12. When it is repaired, S "C" and PI "D" will trigger their respective 8230
to unwrap RO and RI, thus finally completing all the ring paths.
5.1.1.2.5 Summary: The IBM 8230 Controlled Access Unit, with its associated
Lobe Access Modules provides advanced capability over the unpowered IBM 8228
Multi-Station Access Unit.
In conjunction with the IBM LAN Station Manager V1.0, and the IBM LAN Network
Manager, it enables secure asset and access protection, such that unauthorized
stations do not connect to the token-ring.
By means of its internal circuitry and intelligence, it is able to provide
high levels of ring recovery for situations that, without the 8230, would
normally require manual intervention. It is also able to accomplish ring
re-configuration at speed, hence reducing the implications of fault isolation
on the applications that are running on LAN-attached stations.
Because of the pluggability of the ring-in and ring-out modules, the IBM 8230
is adaptable to the changing cabling needs of the customer.
5.1.1.2.6 Device Attachment Cards: Any device attachment hardware component,
for any IBM token-ring attachable device, is complemented by a software and/or
microcode component. The combined components implement the IEEE 802.5 and
IEEE 802.2 LAN standards and optionally, additional LAN management function.
Whether all of the LLC level code is supplied in the software component or as
microcode on a particular token-ring adapter card depends on the type of
attaching device.
5.1.1.3 Repeaters and Converters
The distance travelled by a token before regeneration of the signal is limited
by the attenuation characteristics of the medium. The adapters of active
token-ring devices provide regeneration whether or not the device has data of
its own to send. In some cases the signal must be maintained over longer
stretches of cable where stations cannot be placed. In these cases a special
device called a REPEATER may be used to provide the regeneration.
Only copper wire without repeaters may be used in a LOBE, that is the wire
between the LAN station in the office and a multistation access unit port in
the wiring closet.
When longer distances are encountered between wiring closets (INTER-WIRING
CLOSET CONNECTIONS), active repeaters or converters may be applied. They are
used to amplify and re-synchronize the signal, thus extending the maximum
distance allowed between multistation access units in remote and wiring
closets.
The distance by which a repeater extends the transmitted signal is measured
from the repeater to the next signal driving point. This could be another
repeater (as they can be connected in series) or a ring station.
IBM offers three stand-alone repeaters/converters to provide signal
regeneration over copper or fiber cables between wiring closets, as well as
the IBM 8230 Controlled Access Unit which has integrated repeater/converter
function.
A major consideration in locating repeaters or converters is the distance the
signal must be driven in the event cable on the main path fails or is removed,
since the backup path, which is usually considerably longer than the main
path, will then be used.
For copper wiring between wiring closets, 8218 Copper Repeaters or the IBM
8230 Controlled Access Unit may be installed to provide the necessary signal
regeneration. For several reasons a fiber cable may be used between wiring
closets, for which a pair of 8219 Optical Fiber Repeaters, 8220 Optical Fiber
Converters or the IBM 8230 Controlled Access Unit may be installed to convert
electrical signals to/from light pulses.
A pair of IBM 8220 Fiber Converters or the IBM 8230 Controlled Access Unit
actively participate in the token passing protocol, thereby supporting
automatic monitoring of both paths and automatic recovery from some types of
error. The other repeater devices, the IBM 8218 Copper Repeater and the IBM
8219 Fiber Repeater, are simpler components of the physical cabling and do not
implement MAC layer functions.
Repeaters and converters contribute to accumulated phase jitter(13 under
heading "5.1.1.3 Repeaters and Converters") since they amplify and regenerate
the signal as any LAN station does.
---Footnote---
(13) Every time a signal is amplified and regenerated, a small amount of
additional phase distortion is introduced. This accumulated effect of phase
distortion is referred to as phase jitter.
--------------
Therefore, repeaters and converters must be included when considering the
maximum number of devices supported for any given cable type on a single ring.
5.1.1.4 IBM 8218 Copper Repeater
Figure 54 under heading "5.1.1.4 IBM 8218 Copper Repeater" represents a
Copper Repeater (8218). It is an active component used to cover longer
copper-wired distances between wiring closets.
+------------------------------------------------------------------------------
]
]
]
] *-------------------*
] ] ]
] ] o POWER ON ]
] ] o DATA ERROR ]
] ] ]
] ] *-------* RI ]
] ] ] ) ] ]
] ] ] ) ] ]
] -----------> ) ] ]
] -----------> ) ] ]
] ] *-------* ]
] ] ]
] ] *-------* RO ]
] ] ] ( ] ]
] ] ] ( ] ]
] ] ] ( -------------->
] ] ] ( -------------->
] ] *-------* ]
] ] ]
] *-------------------*
]
]
+------------------------------------------------------------------------------
Figure 54. The 8218 Copper Repeater
The IBM 8218 Copper Repeater operates only at 4 Mbps. When used with Type 1
cable, the maximum distance between two Wiring Closets can be increased to 770
meters. With Type 9 cable, the maximum distance is about 500 meters.
Intermediate 8228 Multistation Access Units are allowed between pairs of 8218
Copper Repeaters. The actual placement of 8218s depends upon the result of
design guidelines, documented in IBM TOKEN-RING NETWORK INTRODUCTION AND
PLANNING GUIDE. This manual also contains the rules and tables required to
plan and verify a physical token-ring installation based on IBM Cabling System
components.
In the 8218, only the main path entering the ring in (RI) connector is
amplified and retimed. Therefore 8218s must be installed in multiple pairs
(minimum four 8218s) to provide signal symmetry and recovery capability via
the backup path.
A sample illustration of 8218 Copper Repeaters without intermediate 8228
Multistation Access Units, Figure 55 under heading "5.1.1.4 IBM 8218 Copper
Repeater", shows why two pairs of 8218s are required to maintain full backup
path capability.
+------------------------------------------------------------------------------
]
]
] oooooooooooooo//ooooooooooooooo
] Patch Panel o*------------//-------------*o Patch Panel
] *-------------o]-* *-]o-------------*
] ] o] ] ] ]o ]
] ] **----* *----** ]
] ] **oooo] ]oooo** ]
] ] ] o] ]o ] ]
] ] ] o] ]o ] ]
] ] ] o] ]o ] ]
] ] ] o] ]o ] ]
] ] ] o] ]o ] ]
] *----------------* \/(S3) (S2)\/ *----------------*
] 8228 #1 /\ /\ 8228 #2
] *----------------* ]o o] *----------------*
] ] RO **] ]o o] ]** RI ]
] ] **] ]o o] ]** ]
] *--------------]o* ]o o] *o]--------------*
] *----------------*o ]o o] o*---------*
] ]oooooooooooooooooo ]o o] ooooooooooo]
] ]o ]o o] (S1) o]
] ]o ]o o] *--\/--* o]
] ]o 8218's ]o o] ]*-/\-*] o]
] ]o(R1) (R4) ]o o] ]] ]] o]
] ]o*--* *--* ]o o] *]]* *]]* o]
] ]oo**]RI]**----------*o o] ]**]RI]**] o]
] *--**] ]**oooooooooooo o] ]**] ]**] o]
] ] ] ] ] o] ] ] ] ] o]
] ]**]RO]**] o*--**]RO]**ooo]
] ]**] ]**] oooo**] ]**---*
] *]o* *]o* *--* *--*
] ]o(S4)]o (R3) (R2)
] ]oo\/-*o 8218's
] *--/\ooo
]
] oo\/-- yellow cross-over
] --/\oo patch cable
] ------ primary path (red/green)
] oooooo backup path (orange/black)
+------------------------------------------------------------------------------
Figure 55. 8218 Copper Repeater Installation
For convenience, the 8218 Copper Repeaters have been identified (from left to
right) as: R1, R4, R3, R2. Cross-over patch cables have been numbered S1, S2,
S3, S4 (from right to left).
The primary path leaving the multistation access unit ring-out on the left
side is normally amplified by R1. Repeater R2 is required at the other end
for signal symmetry.
To provide proper amplification of the backup path signal by R3, a cross-over
patch cable (S1) is used to swap the ring in and ring out wire pairs entering
R2 (only the primary pair (red/green) entering the 8218 Ring-In would normally
be amplified). In this way, the signal on the backup path will be amplified
by Repeater R3.
Before the signal travels back between the wiring closets, the primary and
backup signals must be restored (on the red/green and orange/black pairs
respectively) by cross-over patch cable S2.
The amplification of the backup signal by R3 requires another repeater (R4) at
the other end. At the Ring In of R4 the backup signal must arrive on the
red/green pair. Therefore cross-over cable S3 is required.
Finally, after leaving R4, the backup signal is put back on the orange/black
pair by cross-over cable S4.
5.1.1.5 IBM 8219 Optical Fiber Repeater
The IBM 8219 Optical Fiber Repeater extends the effective drive distance
between wiring closets up to two kilometers using Type 5 cable (containing two
100/140 micron optical fibers). Each IBM 8219 provides a standard 4-wire data
connector for the incoming or outgoing copper cable and two optical connectors
(marked "receive" and "transmit") for the optical fiber connections as
illustrated in Figure 56 under heading "5.1.1.5 IBM 8219 Optical Fiber
Repeater".
+------------------------------------------------------------------------------
]
]
]
] *-------------------*
] ] ]
] ] o POWER ON ]
] ] o LOW SIGNAL ]
] ] o LOW LIGHT ]
] ] ]
] ] ]
] ] o RECEIVE ]
] ] ]
] ] o TRANSMIT ]
] ] ]
] ] ]
] ] *------* ]
] ] ] () ] ]
] ] ] () ] ]
] ] ] () ] ]
] ] ] () ] ]
] ] *------* ]
] ] ]
] *-------------------*
]
]
]
+------------------------------------------------------------------------------
Figure 56. 8219 Optical Fiber Repeater
Other fiber sizes supported are 50/125, 62.5/125 and 85/125. Like the IBM
8218 Copper Repeater, the 8219 operates only at 4 Mbps.
An important advantage of installing fiber cable between wiring closets is
that the optical signal is virtually immune from outside interference. Optical
media should be considered for outdoor connections or where cabling must pass
through electrically noisy areas.
Use of fiber may also be a consideration to simplify coexistence with and/or
migration to FDDI LANs in installations with foreseeable high bandwidth
requirements.
For more information on the use and installation of fiber optic cabling with
current IBM products, refer to IBM TOKEN-RING NETWORK - OPTICAL FIBER CABLE
OPTIONS
Only the signal carried in the primary path (the red/green wire pair in the
data connector) is transmitted via the transmit connector of one 8219 and
received at the receive connector of the other 8219. The optical signal
entering the receive connector is converted back to electrical pulses on the
backup copper wire pair (orange/black).
These considerations are illustrated in the 8219 Fiber Repeater installation
in Figure 57 under heading "5.1.1.5 IBM 8219 Optical Fiber Repeater".
+------------------------------------------------------------------------------
]
]
] *-------------//--------------*
] Patch Panel ] inter-closet fiber ] Patch Panel
] *-------------]--* *--]-------------*
] ] *-* ] <-- ] *-* ]
] ] ]o-----* ---- orange --- *-----o] ]
] ] ]o-----] ---- black ---- ]-----o] ]
] ] *-* ] ] --> ] ] *-* ]
] ] ] ] ] ] ]
] ] ] ] ] ] ]
] ] ] ] ] ] ]
] ] ] ] ] ] ]
] *----------------* ] ] *----------------*
] 8228 ] ] 8228
] *----------------* ] ] *----------------*
] ] RO **] ] ] ]** RI ]
] ] **] ] ] ]** ]
] *--------------]]* ] ] *]]--------------*
] *----------------*] ] ] ]*---\/-----*
] ]*----------------* ] ] *----/\----*]
] ]] ] ] ]] Yellow
] ]] ] ] ]] cross-
] ]] 8219 ] ] 8219 ]] over
] ]] ] ] ]] patch
] ]] *--* ] ] *--* ]] cable
] ]] ]Ro--* <-orange ] ]black->*--oR] ]]
] ]] ] ] ]-------------* *-------] ] ] ]]
] ]] ]To--* black-> <-orange*--oT] ]]
] ]] ] ] ] ] ]]
] ]*--**] ]**--*]
] *---**] ]**---*
] *--* *--*
]
]
]
+------------------------------------------------------------------------------
Figure 57. 8219 Optical Fiber Repeater Installation
Because they use only copper connectors, intermediate 8228 Multistation Access
Units cannot be positioned between paired 8219 Optical Fiber Repeaters.
This diagram also shows that only one pair of 8219 Fiber Repeaters is required
to provide a connection between wiring closets with full backup capabilities.
Note that only one yellow cross-over patch cable is required compared to the
four cables required by the 8218 example described earlier.
When the primary signal enters the left 8219 on the red/green copper pair, it
will be transmitted as an optical signal on the black fiber. On the right
side, the black fiber is connected to the receive connector of the second 8219
where the signal is converted and sent out on the orange/black copper pair.
Therefore, after leaving the right 8219, a cross-over cable is applied to put
the primary signal back on the red/green copper pair.
In backup situations, the signal will be wrapped at the two 8228 access units.
Consequently the cross-over cable will swap the backup signal to the red/green
copper wire pair as it leaves the ring-in of the right multistation access
unit. Since the backup signal enters the right 8219 on the red/green pair, it
will be transmitted as an optical signal on the orange fiber. On the left
side, the orange fiber enters the receive connector of the 8219. The optical
light pulses will be converted back to electrical signals and sent out on the
orange/black copper pair. This direction is correct for the backup signal as
it enters the ring-out of the left multistation access unit.
5.1.1.6 IBM 8220 Optical Fiber Converter
The 8220 Optical Fiber Converter operates in a token-passing ring to extend
the distance between multistation access units in separate wiring closets up
to two kilometers when operating at either 16 Mbps or at 4 Mbps.
The 8220 Optical Fiber Converter, unlike the 8218 Copper Repeater or 8219
Optical Fiber Repeater, participates in the token passing ring protocols,
enabling it to provide error detection and automatic recovery functions in
some circumstances. As this implies, each 8220 Optical Fiber Converter has a
unique individual MAC address to support this capability.
The 8220 Fiber Optical Repeater has one token-ring media data connector to
connect the 8220 to the ring in or ring out position of an 8228 Multistation
Access Unit, two optical connectors (marked "receive" and "transmit") as shown
in Figure 58 under heading "5.1.1.6 IBM 8220 Optical Fiber Converter" to
connect to the fiber cables.
+------------------------------------------------------------------------------
]
]
]
] *-------------------*
] ] ]
] ] o Power On ]
] ] o Insert ]
] ] o E1 Error ]
] ] o E2 Error ]
] ] ]
] ] ! 4 / 16 Mbps ]
] ] ! RI / RO ]
] ] ]
] ] o RECEIVE ]
] ] ]
] ] o TRANSMIT ]
] ] ]
] ] *------* ]
] ] ] () ] ]
] ] ] () ] ]
] ] ] () ] ]
] ] ] () ] ]
] ] *------* ]
] ] ]
] *-------------------*
]
]
]
]o The Insert LED is on when the 8220 is inserted.
]o E1 and E2 Error LEDs indicate internal and external errors
] respectively.
]o The 4 Mbps/16 Mbps switch enables the 8220 to operate at the
] installation-specified rate.
]o The Ring-In/Ring-Out switch indicates the 8220 operating mode as
] either Upstream (RI) or Downstream (RO).
+------------------------------------------------------------------------------
Figure 58. 8220 Optical Fiber Converter
Apart from offering two operational data rates, 4 MBPS and 16 MBPS, the 8220
Optical Fiber Converter also has two operational states, WRAP state and
INSERTED state.
o In "wrap" state, the 8220 is not operational. It is physically and
logically NOT-INSERTED into the ring. Due to internal wrapping, a
fiber-connected pair of 8220 Optical Fiber Converters form a "closed
ring", allowing self-test diagnostics to execute automatically. IBM 8220
Fiber Converters can be removed from and reinserted into a ring like other
token-ring stations.
o In "inserted" state, the 8220 is physically and logically INSERTED in both
the main and backup paths of the ring, transmitting additional 8220 unique
frames as well as IEEE 802.5 MAC PROTOCOL (non-early-token-release)
frames.
The 8220 Optical Fiber Converter can, by means of the RI/RO switch, be set
into either of the following operational modes: UPSTREAM (Ring-In) or
DOWNSTREAM (Ring-Out).
A pair of IBM 8220 Optical Fiber Converters connected as shown in diagram
Figure 59 under heading "5.1.1.6 IBM 8220 Optical Fiber Converter" is called
a 8220 FIBER OPTICAL CONVERTER SUBSYSTEM. Both converters are referred to as
PARTNERS within the 8220 Subsystem protocols.
-------------------------------------------------------------------------------
8228/8230 8228/8230
Access Access
Unit Unit
---------* *---------
] ]
RO] ] Upstream Downstream ] ]RI
] ] *------* *------* ] ]
-------+-* ] 8220 ] ] 8220 ] *-+-------
] ] ] ] ] ]
] ] Tx]-------------]Rx ] ]
] ] ] ] ] ]
*-----] Rx]-------------]Tx ]-----*
A ] ] ] ] A
] *------* *------* ]
Normal---* Switched Switched *---Yellow
Black to RI to RO X-over
Patch cable cable
-------------------------------------------------------------------------------
Figure 59. Cabling an IBM 8220
In this diagram:
o The upstream (left) converter is connected to the ring-out position of an
8228 Multistation Access Unit. Logically, the upstream converter
participates in the ring traffic carried in the main path.
Inversely, the downstream converter is connected to the ring-in position
of the next 8228 Multistation Access Unit. Logically, the downstream
converter participates in the ring traffic carried in the backup path.
o The upstream converter provides the control function within the 8220
subsystem; the downstream converter follows the control function provided
by the upstream converter.
o If a loss of power or failure occurs within the scope of the 8220
subsystem, the 8220 pair will automatically de-insert from the ring and
wrap the main ring to the backup ring at each end of the subsystem, thus
maintaining the operation of the ring. This is shown schematically in
Figure 60 under heading "5.1.1.6 IBM 8220 Optical Fiber Converter". As
soon as the optical fiber cable is repaired, the 8220 subsystem will
return to normal operation.
o An 8220 Optical Fiber Converter can be defined as a CRITICAL RESOURCE to
the IBM LAN Manager V2 or IBM LAN Network Manager
o The downstream 8220 continuously monitors the condition of the backup ring
so that a potential failure in the backup ring can be identified and
corrected by operations personnel before the ring is needed for backup
operation. This is accomplished by the adapter on the backup path sending
a Downstream Converter Presence LLC frame to the functional address of LAN
Manager every 60 seconds.
+------------------------------------------------------------------------------
]
]
] Primary Ring
]*-----------------------------------<----------------------------------*
]] *--------------------------------->--------------------------------* ]
]] ] Backup Ring ] ]
]] ] ] ]
]] ] *----------------------**---* fiber *---**---------------------* ] ]
]] ] ]RI MSAU 1 RO]] ] cable ] ]]RI MSAU 2 RO] ] ]
]] ] ]*---------<----------*]] *----//-----* ]]*---------<---------*] ] ]
]] ] ]]***------>---------*]]] *----//-----* ]]]***------>------***]] ] ]
]] ] *]]]]--------**------]]*] ] break ] ]*]]]]---**--------]]]]* ] ]
]] *--*]]] ]] ]] ]RI ] ] RO] ]]]] ]] ]]]*--* ]
]*-----*]] ]] ]*-<-* ] ] *-<-*]]] ]] ]]*-----*
] ]] ]] *-->-* ] ] *->--*]] ]] ]]
] *--* *--* *---* *---* *--* *--* *--*
] ]S1] ]S2] ]S3] ]S4] ]S5]
] *--* *--* 8220 8220 *--* *--* *--*
] Upstream Downstream
]
]
+------------------------------------------------------------------------------
Figure 60. 8220 Subsystem - Automatic Wrap to Backup Ring
For additional installation considerations and the use of cross-over patch
cables, refer to the 8219 installation shown in Figure 57 under heading
"5.1.1.5 IBM 8219 Optical Fiber Repeater".
5.1.1.6.1 Summary: The IBM Cabling System offers cable types and components
to implement general purpose structured wiring for data communications. It
also allows, within specified limitations, use of existing telephone twisted
pair wire for transmission of digital data.
The structured wiring approach implemented by the IBM Cabling System very well
suits the cabling requirements of the IBM Token-Ring Network, both at 4 Mbps
and 16 Mbps data rates. Unshielded telephone twisted pair wire is only
applicable to 4 Mbps token-ring cabling. Between multistation access units in
distant wiring closets, optical fiber cable may be installed.
Of the three repeaters/converters offered to extend the distance between
multistation access units in separate wiring closets, the 8220 Optical Fiber
Converter may frequently provide the most desirable solution. It is the only
repeater/converter device supporting both 4 Mbps and 16 Mbps data rates
between wiring closets, and is currently the only device that implements
automatic recovery for cable failures between wiring closets (since it
actively participates in the IEEE 802.5 MAC protocols). The IBM 8230
Controlled Access Unit, with its integrated repeaters/converters changes the
picture, and, because it is an active device, can offer superior function and
recovery capability.
5.1.2 IBM Token-Ring Network Devices
5.1.2.1.1 Workstation Adapters: The original IBM Token-Ring Network Adapter,
announced in 1985 and often referred to as Token-Ring Adapter I, has been
withdrawn from marketing.
IBM Personal Computers and the IBM Personal System/2 Model 30, subsequently
referred to as Family 1 workstations, can be attached to the IBM Token-Ring
Network through the IBM TOKEN-RING NETWORK ADAPTER II at a data rate of 4
Mbps. The IBM 16/4 TOKEN-RING NETWORK ADAPTER offers a 16 Mbps / 4 Mbps
switchable token-ring attachment for the same Family 1 devices. This adapter
has now been superseded by the IBM 16/4 TOKEN-RING NETWORK ADAPTER ENHANCED.
All other IBM Personal System/2 models, equipped with the Micro Channel bus
and subsequently referred to as Family 2 workstations, can be attached to the
IBM Token-Ring Network through the IBM TOKEN-RING NETWORK ADAPTER/A at a data
rate of 4 Mbps or through the IBM 16/4 TOKEN-RING NETWORK ADAPTER/A to either
a 16 Mbps or 4 Mbps token-ring LAN. As with the Family 1 workstations, this
16/4 switchable adapter has been superseded by the IBM 16/4 TOKEN-RING NETWORK
ADAPTER ENHANCED.
When attached to a 16 Mbps token-ring LAN using the proper Family 1 or Family
2 adapter card, the IEEE 802.5 protocol implementation provides the early
token release option.
Special token-ring adapters are available for both Family 1 and Family 2 PCs
and PS/2s (14 under heading "5.1.2.1.1 Workstation Adapters")
---Footnote---
(14) PC and PS/2 are registered trademarks of the International Business
Machines Corporation.
--------------
to support the IBM Token-Ring Network Trace and Performance Program,
respectively the IBM 16/4 TOKEN-RING NETWORK TRACE AND PERFORMANCE ADAPTER and
the IBM 16/4 TOKEN-RING NETWORK TRACE AND PERFORMANCE ADAPTER /A. The IBM
Token-Ring Network Trace and Performance facilities are discussed in "9.7 IBM
Token-Ring Network Trace and Performance Facilities."
The RT/PC provides also a 4 Mbps attachment to an IEEE 802.5 LAN.
5.1.2.1.2 Adapter Support Interface for Token-Ring Attached Workstations:
IBM LAN SUPPORT PROGRAM: The adapter support interface between the
workstation's processor and any token-ring adapter card is included in the IBM
LAN SUPPORT PROGRAM. More details on the other functions of this product are
provided in "5.6 IBM LAN Support Program." One significant function with
respect to adapter support however is to provide access to memory locations on
a token-ring adapter card from which the workstation processor can retrieve
received information or to which it must place data for transmission. These
memory locations reside in an area of the adapter card called SHARED RAM.
All currently offered 4 Mbps IBM Token-Ring Network adapters have a Shared RAM
of 16 Kbytes. The 16/4 Mbps switchable adapters have a shared RAM of 64 KB
which can be divided into four 16 KB pages. All or part of these memory pages
may be used by system software. A technique called RAM PAGING, supported by
the IBM LAN Support Program, permits use of the entire 16 or 64 KB shared RAM
memory while requiring only 16 KB in the PC or PS/2 memory.
+----------------------------------------------------------------------------+
] ]
] PS/2 Memory ]
] *----------------------* ]
] ] ] ]
] 64 KB Shared RAM / / ]
] *----------------------* ] ] ]
] ] 512 Bytes - reserved ] ] ] ]
] ]----------------------] ]----------------------] --------- ]
] ] ] ] ] ] ]
] ] 15.5 Kbytes - Page 4 ] <---* ] ] ] ]
] ] ] ] ] ] ] ]
] ]----------------------] ] ]----------------------] ] ]
] ] ] ] ]///// /] 128 KB ]
] ] 16 Kbytes - Page 3 ] <---+----> ]/// 16 Kbytes ///] adapter ]
] ] ] ] ]/ /////] mapping ]
] ]----------------------] ] ]----------------------] ] ]
] ] ] ] ] ] ] ]
] ] 16 Kbytes - Page 2 ] <---] ] ] ] ]
] ] ] ] ] ] ] ]
] ]----------------------* ] ]----------------------* --------- ]
] ] (8 KB - 1B) ] ] ] ] ]
] ]- 16 Kbytes - Page 1 -] <---* ] ] ]
] ] (8 KB - 1A) ] / / ]
] *----------------------* ] ] ]
] *----------------------* ]
] ]
] ]
+----------------------------------------------------------------------------+
Figure 61. RAM Paging
In general, a larger shared RAM size allows larger frames to be transmitted,
(up to a maximum 4 KB frame size on a 4 Mbps token-ring adapter with 16 KB
shared RAM versus 16 KB maximum frame size on a 16/4 Mbps token-ring adapter
with 64 KB shared RAM. Use of larger frame sizes provides the potential for
improved ring utilization. Larger shared RAM memory size also enables
support for more (up to 254) link stations (see "3.2.2 Logical Link Control
Sublayer").
Figure 62 under heading "5.1.2.1.2 Adapter Support Interface for Token-Ring
Attached Workstations" shows the default shared RAM utilization as implemented
by the IBM LAN Support Program for a 16 KB shared RAM size.
+----------------------------------------------------------------------------+
] ]
] ]
] Ring Station A Ring Station B ]
] ]
] 16 KB Shared RAM 16 KB Shared RAM ]
]*------------------------* *------------------------* ]
]] Work Area - 1588 bytes ] ] Work Area - 1588 bytes ] ]
]]------------------------] ]------------------------] ]
]] SAPs n x 64 bytes ] ] SAPs n x 64 bytes ] ]
]]------------------------] ]------------------------] ]
]] Link Stations ] ] Link Stations ] ]
]] m x 144 bytes ] ] m x 144 bytes ] ]
]]------------------------] ]------------------------] ]
]] Transmit Buffers ] ] Transmit Buffers ] ]
]]------------------------] Frame (p bytes) ]------------------------] ]
]] Receive Buffers ] ---------------> ] Receive Buffers ] ]
]] ] <--------------- ] ] ]
]] - - - - - - - - - - - -] Frame (q bytes) ] - - - - - - - - - - - -] ]
]] unused ] ] unused ] ]
]*------------------------* *------------------------* ]
] ]
] Receive Buffers Transmit Buffers ]
] <--- r bytes ---> <------- t bytes --------> ]
] ] ] ] ] ]
] *-----------------* *------------------------* ]
] ] ]
] *>] ] ]
] *-----------------* Number of ]Transmit Buffers]Receive Buffer ]
] ] Link Stations]Number Size] Size ]
] *>] ] -------------+----------------+-------------- ]
] *-----------------* 1 - 32 ] 2 2040] 280 ]
] ] 33 - 48 ] 2 1048] 280 ]
] *>................. 49 - 64 ] 1 1048] 280 ]
] *- > 64 ] 1 600] 144 ]
] ] ]
] *<] ] ]
] *---------------* ]
] ]
+----------------------------------------------------------------------------+
Figure 62. Shared RAM Utilization
The maximum frame length is determined as the smallest of either the
transmission buffer size or half the number of receive buffers times the
receive buffer size. The number of receive buffers is a function of the
number of SAPs and the number of link stations. A realistic maximum number of
link stations is 64 for a token-ring adapter with 16KB RAM.
Larger frame sizes can be obtained by reducing the number of required SAPs and
link stations. Default parameters supplied by the IBM LAN Support Program
should be evaluated with respect to the LAN applications used by the
workstation to ensure that neither too few or too many link stations, SAPs,
sessions, and outstanding commands are defined.
Receive Buffers are chained to accommodate at least twice the transmit frame
size. This prevents overrun for incoming frames received faster than can be
handled by the receiving adapter and processor.
Currently IBM Token-Ring Network adapters can be equipped with a REMOTE
INITIAL PROGRAM LOAD (RIPL) feature, consisting of a pluggable EPROM. This
feature allows a token-ring attached device to obtain a bootstrap program from
a remote LAN RIPL server at power-on time, rather than supplying its own
startup code. For additional information, see "8.1 IBM PC LAN Program V1.2."
5.1.3 IBM PC Network (Broadband) Components
The IBM PC Network (Broadband) consists of a set of IBM products that allow
the interconnection of IBM PCs to form a local area network in a tree
topology.
The IBM PC Network (Broadband) components are:
o IBM PC Network Adapter cards
o IBM PC Network Translator Unit and Splitter or other commercial
translators used with customer cabling
o IBM PC Network (Broadband) cabling components:
- IBM Base Expander
- IBM Short Distance Kit
- IBM Medium Distance Kit
- IBM Long Distance Kit
o Adapter Support software (the IBM LAN Support Program).
Figure 63 under heading "5.1.3 IBM PC Network (Broadband) Components" shows
the structure of the IBM PC Network (Broadband) local area network. Using IBM
components the network can support up to 72 workstations, at a distance of up
to 305 meters (1000 ft.) from the translator unit.
+------------------------------------------------------------------------------
]
]
] 8 nodes direct
] *-----* 8-way *-----* short
] ]oooo ] splitter ]oooo ] distance
] *-----] oooo] *-----] oooo] kit
] ] *-----* ] *-----*
]*------* ] 0.3 m ] *-----*
]] ] *-* *-------------* 122 meters ]oooo ]
]] ]---]C]---]oooo ]-----------------------------] oooo]
]] ] *-* ] oooo]-------* medium distance kit *-----*
]*------* *-----* ] *-----*
]Translator Base ] 244 meters ]oooo ]
]Unit Expander *----------------------------------] oooo]
] long distance kit *-----*
]C = directional coupler
]kit-to-node: max. 61 m
]
] *------------------------------------------------*
] ] IBM cable ] Custom Network ] Custom Network ]
] ] kits and IBM ] and IBM ] and commercial ]
] ] translator ] translator ] translator ]
]*-----------------+--------------+----------------+----------------]
]]multiple channels] no ] no ] yes ]
]]distance (radius)] 305 m ] 305 m ] 5 Km ]
]]number of nodes ] 72 ] 256 ] 1000 ]
]*------------------------------------------------------------------*
]
+------------------------------------------------------------------------------
Figure 63. IBM PC Network (Broadband) Components
The number of PCs that can be attached to the network, and the distance that
they can be from the translator unit can be increased by using non-IBM
equipment. For example, using the IBM translator and non-IBM expander(s) and
cabling, 256 PCs can be supported. Using a third-party translator, up to
1000 workstations may be attached while the network radius may be increased to
about 5 kilometers.
5.1.3.1.1 PC Network (Broadband) Adapters: PCs or PS/2s are attached to the
network through a PC Network (Broadband) Adapter card. Several adapters are
available: the original IBM PC Network Adapter and a number of second
generation adapters including the IBM PC Network Adapter II for Family 1
devices, the IBM PC Network Adapter II/A for Family 2 devices, and PC Network
Adapters for both families of devices operating at different transmit and
receive channels (Frequency 2 and Frequency 3 adapters).
IBM PC NETWORK ADAPTER
The original PC Network adapter fits only Family 1 devices. It uses
Radio Frequency (RF) channels T13 and J from the mid-split channel
range (offering 17 channel pairs, with a 168.25 MHz offset).
Card-mounted Intel 82586 and 80188 processors provide the CSMA/CD
support as well as support for a higher level protocol, NETBIOS, in
ROM on the adapter.
IBM PC NETWORK ADAPTER II
This second generation adapter for Family 1 devices uses the same RF
channels T13 and J as the original PC Network adapter. The adapter
uses an Intel 82588 LAN controller chip to provide framing and media
access functions, but does not provide any higher-layer protocol in
ROM.
IBM PC NETWORK ADAPTER II/A
This adapter is equivalent to the PC Network Adapter II but is
designed for Family 2 devices with smaller card forms and Micro
Channel support.
IBM PC NETWORK ADAPTER II, II/A FREQUENCY 2
These adapters for Family 1 and Family 2 devices respectively offer
the same features as the IBM PC Network Adapters II and II/A, except
that they transmit and receive on different channels frequencies.
Frequency 2 Adapters use RF channels 2' and O from the high-split
broadband range (30 channel pairs with a 192.25 MHz offset).
High-split is commonly used in a manufacturing environment, where
these adapters can coexist (not communicate) with other devices
attached to the same broadband medium, for example, MAP 2.1
adapters.
IBM PC NETWORK ADAPTER II, II/A FREQUENCY 3
Again these adapters are functionally equivalent to the other second
generation (II) PC Network adapters, except that these adapters
operate on high-split RF channels 3' and P.
Figure 64 under heading "5.1.3.1.1 PC Network (Broadband) Adapters" lists the
part numbers of all current PC Network (Broadband) adapters.
+----------------------------------------------------------------------------+
] ]
] ]
]*----------------------------------------------------------------------* ]
]] IBM PC Network (Broadband) Adapters ] Part / Feat.] 04/1989 ] ]
]] (AFE) ] Number / Code ] Part No.] ]
]]-------------------------------------------+----------------+---------] ]
]] IBM PC Network Adapter ] 6450213 / 0213 ] N/A ] ]
]] ] ] ] ]
]] IBM PC Network Adapter II ] 1501220 / 1220 ] 25F8279 ] ]
]] IBM PC Network Adapter II - Frequency 2 ] 96X5645 / 5645 ] 25F8282 ] ]
]] IBM PC Network Adapter II - Frequency 3 ] 96X5646 / 5646 ] 25F8285 ] ]
]] ] ] ] ]
]] IBM PC Network Adapter II/A ] 1501222 / 1222 ] 25F8278 ] ]
]] IBM PC Network Adapter II/A - Frequency 2 ] 96X5647 / 5647 ] 25F8281 ] ]
]] IBM PC Network Adapter II/A - Frequency 3 ] 96X5648 / 5648 ] 25F8284 ] ]
]*----------------------------------------------------------------------* ]
] ]
]*----------------------------------------------------------------------* ]
]] IBM PC Network (Broadband) Adapters ] Part / Feat.] 04/1989 ] ]
]] (EMEA) ] Number / Code ] Part No.] ]
]]-------------------------------------------+----------------+---------] ]
]] IBM PC Network Adapter ] 6450213 / 0213 ] N/A ] ]
]] ] ] ] ]
]] IBM PC Network Adapter II ] 1501220 / 1220 ] ] ]
]] IBM PC Network Adapter II - Frequency 2 ] 96X5645 / 5645 ] ] ]
]] IBM PC Network Adapter II - Frequency 3 ] 96X5646 / 5646 ] ] ]
]] ] ] ] ]
]] IBM PC Network Adapter II/A ] 1501222 / 1222 ] 25F8280 ] ]
]] IBM PC Network Adapter II/A - Frequency 2 ] 96X5647 / 5647 ] 25F8283 ] ]
]] IBM PC Network Adapter II/A - Frequency 3 ] 96X5648 / 5648 ] 25F8286 ] ]
]*----------------------------------------------------------------------* ]
+----------------------------------------------------------------------------+
Figure 64. IBM PC Network (Broadband) Adapters
There is no communications path between PC Network (Broadband) adapters
operating on different channel pairs, even if they are attached to the same
common medium. All adapters operating on the same transmit and receive
channels form one LAN segment, physically and logically separated from
adapters operating at a different channel pair. Connectivity can only be
provided through LAN segment interconnection such as the PC Network Bridge
Program, see "6.6.3 IBM PC Network Bridge Program" for more details.
All second generation PC Network Adapters require the IBM LAN SUPPORT PROGRAM
to provide the adapter support interface and make the LAN protocols accessible
to applications. The original IBM PC Network Adapter, having NETBIOS in Read
Only Memory (ROM) on the adapter card, contains its own interface and gives
access to the LAN for application programs through NETBIOS. However, not all
NETBIOS commands are supported and no other application interfaces are
supported. The IEEE 802.2 LLC interface or other higher layer protocols (for
example, SNA protocols) are not available unless the IBM LAN Support Program
is used to bypass the ROM NETBIOS, and provide the same interfaces as for the
second generation adapters. A more detailed discussion about LAN Support
Program and coexistence considerations is presented in "5.6 IBM LAN Support
Program."
5.1.3.1.2 IBM Translator Unit and Splitter: The IBM PC Network (Broadband)
Translator Unit provides broadband frequency translation for a passive IBM PC
Network (Broadband). One translator is required for each network. The IBM
Translator Unit provides only single channel frequency shift between modulated
signals transmitted at a 50.75 MHz center frequency and a receive channel with
a 219 MHz center frequency. Only this mid-split channel pair is recognized by
the IBM Translator Unit. All other signals are ignored. Third party
translators are required to operate a PC Network (Broadband) LAN at the
high-split frequencies used by the PC Network (Broadband) Frequency 2 or 3
adapters.
The IBM PC Network Translator Unit is packaged together with a directional
coupler and an 8-way splitter, supporting direct connection for eight nodes up
to 61 meters (200 feet) from the splitter. See Figure 63 under heading "5.1.3
IBM PC Network (Broadband) Components" for an illustration of the components.
The translator supports a single channel-pair, data-only network of 256 nodes
if custom cabling is used rather than the pre-cut cabling kits. The nodes (PC
or PS/2 workstations) can be up to 305 meters (1000 feet) away from the
translator unit. Replacement of the IBM Translator Unit by a more
sophisticated device is required if broadband multiple channel capability is
desired. In this case however, the calibration and tuning by a CATV
specialist should planned for both initial installation and on-going
maintenance.
The directional coupler also has a port for attaching an IBM PC Network
(Broadband) Base Expander.
5.1.3.1.3 IBM PC Network (Broadband) Base Expander: The Base Expander allows
for the attachment of up to eight additional cabling kits. Three different
kits are available: the IBM short distance kit, medium distance kit, and long
distance kit.
Workstations cannot be connected directly to the base expander unit. They
must be connected via one of the cabling kits. By using the base expander and
the cabling kits an additional 64 PCs (up to a maximum of 72 workstations) can
be attached to the network.
5.1.3.1.4 Cabling Kits: Because of the sensitivity of the transmitters and
receivers in radio frequency broadband transmission to cables that are either
very short or very long, the use of pre-defined cable lengths minimizes and in
most cases eliminates the effort and cost of recalibrating the frequencies.
o The IBM PC Network (Broadband) Short Distance Kit connects directly to the
base expander and allows up to eight nodes to be connected. Cables are
available with lengths up to 61 meters (200 feet) to connect a PC or PS/2
to a splitter.
o The IBM PC Network (Broadband) Medium Distance Kit allows for the
attachment of a splitter at 122 meters (400 feet) from the base expander.
Hence, the maximum distance a node can be away from the translator is 183
meters (600 feet).
o The IBM PC Network (Broadband) Long Distance Kit consists of a cable that
is 244 meters (800 feet) in length and a splitter. PCs or PS/2's attached
to the splitter can be up to 305 meters (1000 feet) away from the
translator unit.
5.1.4 IBM PC Network (Broadband) Interfaces
5.1.4.1.1 IBM PC Network Medium Access Control: The IBM PC Network Adapter
implements a CSMA/CD protocol like that of the IEEE 802.3 standard, although
it does not provide an IEEE 802.3 interface. Apart from transmission speed,
technique, and network topology used, the second generation PC Network
Adapters (II) are similar to the IEEE 802.3 standard with respect to the
CSMA/CD protocol and support for the IEEE 802.2 standard interface. Therefore
the way stations access the medium, the MAC frame structure, and the interface
provided by the adapters conform to the description of CSMA/CD LANs provided
in "2.2.1 Basic CSMA/CD Concepts."
5.1.4.1.2 IEEE 802.2 Logical Link Control Support: For the second generation
adapter cards, the IEEE 802.2 layer support and interface is provided by the
IBM LAN Support Program. The IBM LAN Support Program optionally provides the
same support and interfaces for the IBM PC Network Adapter as it bypasses and
replaces the NETBIOS ROM with its own software modules.
The IBM LAN Support Program optionally provides a session level NETBIOS
interface, running on top of the IEEE 802.2 LLC module. See "5.6 IBM LAN
Support Program" for more information.
5.1.4.1.3 Higher Layer Protocol Support: NETBIOS is a proprietary
higher-layer protocol providing session connection support on top of the IEEE
802.2 interface through service access point value X'F0.
SNA service access points provide the required interfaces between the PC
Network (Broadband) LAN and SNA path control via service access point value
X'04 or multiple of X'04.
The 8232 LAN Channel Station provides the same connectivity to IBM PC Network
attached workstations as to IBM Token-Ring Network attached workstations, with
one exception. A non-SNA IBM host, running AIX/370, is not accessible from a
PC Network device and this environment is not supported. The 8232 LAN Channel
station may be equipped with any Family 1, second generation PC Network
adapters (Frequency 1, 2 or 3).
5.2 IBM PC Network Baseband
The IBM PC Network Baseband is designed to be a low-cost local area network
for IBM PCs and PS/2's. The network uses baseband transmission at 2 Mbps. The
medium access protocol is, as for IBM PC Network Broadband, CSMA/CD.
The network is made up from a series of segments, each of which is a
daisy-chain of up to eight workstations. If the IBM PC Network Baseband
consists of eight or fewer workstations, the topology can remain as a single
daisy-chain, Figure 65 under heading "5.2 IBM PC Network Baseband" with a
maximum length of 61 meters (200 feet). Such a topology is illustrated in
Figure 65 under heading "5.2 IBM PC Network Baseband" .
+----------------------------------------------------------------------------+
] ]
] ]
] <------------------ 61 meters (8 stations) --------------------> ]
] ]
] *-* *--------------* *--------------* *---------//---------* *-* ]
] ]T] ] ] ] ] ] ] ]W] ]
]*-------* *-------* *-------* *-------* ]
]] ] ] ] ] ] ] ] ]
]] ] ] ] ] ] ] ] ]
]] ] ] ] ] ] ] ] ]
]*-------* *-------* *-------* *-------* ]
] ]
] ]
+----------------------------------------------------------------------------+
Figure 65. IBM PC Network Baseband - Single Segment Daisy-Chain
There is a wrap plug at one end of the chain while the other end is terminated
with a terminator plug.
If more than eight workstations are required in the network the IBM 5173 PC
Network Baseband Extender must be used. With the PC Network Baseband Extender,
up to 80 workstations can be attached to the network in a multiple daisy-chain
configuration, illustrated in Figure 66 under heading "5.2 IBM PC Network
Baseband".
+----------------------------------------------------------------------------+
] ]
] ]
]*-----*<---------------------- 122 meters -------------------------> ]
]] ] ]
]] 5 I]------------* *----------* *----------* *-------//-------* *-* ]
]] 1 I] ] ] ] ] ] ] ] ]T] ]
]] 7 I] *------* *------* *------* *------* ]
]] 3 I] ] ] ] ] ] ] ] ] ]
]] I]----* ] ] ] ] ] ] ] ] ]
]] I] ] ] ] ] ] ] ] ] ] ]
]] E I] ] *------* *------* *------* *------* ]
]] X I] ] ]
]] T I] *-------* *----------* *----------* *-------//-------* *-* ]
]] E I] ] ] ] ] ] ] ] ]T] ]
]] N ]-* *------* *------* *------* *------* ]
]] D O]T] ] ] ] ] ] ] ] ] ]
]] E ]-* ] ] ] ] ] ] ] ] ]
]] R ]-* ] ] ] ] ] ] ] ] ]
]] O]W] *------* *------* *------* *------* ]
]] ]-* ]
]*-----* ]
] ]
] I = IN Ports (10) T = Terminator plug ]
] O = OUT Ports (2) W = Wrap plug ]
] ]
+----------------------------------------------------------------------------+
Figure 66. IBM PC Network Baseband - Multiple Daisy-Chain Segments
If more than 80 workstations are required, then the IBM 5173 Baseband
extenders can themselves be daisy-chained together, up to a maximum of ten.
The largest PC Network Baseband configuration can thus be 800 workstations.
5.2.1 PC Network Baseband Components
5.2.1.1.1 IBM PC Network Baseband Adapters: Two adapters are available for
the IBM PC Network Baseband: the IBM PC Network Baseband Adapter for the
Family 1 workstations, and the IBM PC Network Baseband Adapter/A for Family 2
PS/2s. Instead of a radio frequency modem as on IBM PC Network (Broadband)
adapters, they use a transceiver on the adapter.
The adapters together with the IBM LAN Support Program provide CSMA/CD medium
access control, the IEEE 802.2 LLC protocol and an open interface to the
logical link control services. Because the data rate (2 Mbps), topology and
electrical characteristics deviate from a Standard IEEE 802.3 CSMA/CD LAN
using twisted pair media, the IBM PC Network Baseband does not fully conform
to the international standard.
The adapter cards allow up to eight LAN stations to be daisy-chained together,
with an overall length of up to 61 meters (200 feet). The length of the chain
is dependent on the number of workstations in the chain. For example if there
are only two workstations in the chain (the minimum number) the distance
between the two workstations can be up to 91.4 m (300 feet). Any distance
between two workstations in a chain is allowed as long as the total distance
between the first and last node does not exceed the recommended maxima listed
below:
*------------------------------*
] NUMBER OF ] MAXIMUM ]
] NODES ] DISTANCES ]
]-----------+------------------]
] 2 ] 91.4 m (300 ft) ]
]-----------+------------------]
] 3 ] 83.8 m (275 ft) ]
]-----------+------------------]
] 4 ] 76.0 m (250 ft) ]
]-----------+------------------]
] 5 ] 68.5 m (225 ft) ]
]-----------+------------------]
] 6 ] 68.5 m (225 ft) ]
]-----------+------------------]
] 7 ] 61.0 m (200 ft) ]
]-----------+------------------]
] 8 ] 61.0 m (200 ft) ]
*------------------------------*
Figure 67. Length of a Single Segment Daisy-Chain
In any chain like the example illustrated in Figure 65 under heading "5.2 IBM
PC Network Baseband", the workstation at one end must have a terminator plug
and the station at the other end must have a wrap plug.
The adapter card ROM contains self-test diagnostics that are run when the
adapter is powered on. These diagnostics help a user to identify adapter
failures without a diagnostic diskette or problem determination manuals.
5.2.1.1.2 IBM 5173 PC Network Baseband Extender: The IBM 5173 PC Network
Baseband Extender is used to expand the IBM PC Network Baseband from eight to
eight hundred nodes. The PC Network Baseband Extender has ten ports, each able
to support a chain of maximum eight LAN stations. Each chain attached to the
PC Network Baseband Extender has a maximum length of 121.92 m (400 ft). This
distance is measured from the extender to the last workstation in the chain.
The IBM 5173 PC Network Baseband Extender is designed for continuous,
unattended operation, therefore it does not have an on/off switch. There are
also no option switches or jumpers. The PC Network Baseband Extender has one
indicator light which is illuminated when the power is on. The indicator can
either be red or green, depending on whether an error was detected by the
self-test circuitry. The self-test function is initiated by depressing a
button on the front panel.
The PC Network Baseband Extender contains a 12-watt universal power supply.
It has 10 "IN" ports on the front panel for connecting chains of PCs to the
extender, and two "OUT" ports as shown in Figure 66 under heading "5.2 IBM PC
Network Baseband". The OUT ports are used for daisy-chaining the extenders
together, up to the maximum of ten. A wrap plug must be inserted in an OUT
port of the first extender, and a terminator plug in an OUT port of the last
extender. A terminator plug must also be inserted into the adapter card of
the last workstation of each chain attached to the extender.
Cabling is not supplied with the extender, nor with the PC Network Baseband
Adapter cards, but must be ordered separately.
5.2.1.1.3 PC Network Baseband Cabling and Accessories: The IBM PC Network
Baseband uses twisted pair wiring for interconnecting workstations. Three
adapter cables are offered to provide connector compatibility with different
types of building wiring. Each cable connects to the workstation adapter with
a modular telephone jack (J11). The other end provides either another modular
jack or a different connector type to support daisy chaining of workstations,
or connection wall jacks or sockets.
THE IBM PC NETWORK BASEBAND ADAPTER CABLE (1501229) is shielded twisted pair
cable with modular telephone connectors at each end. It is designed to
serially connect the IBM PC Network Baseband Adapter or the IBM PC Network
Baseband Adapter/A to other workstations in a daisy-chain fashion, or to
connect a workstation to the IBM 5173 PC Network Baseband Extender.
THE IBM PC NETWORK BASEBAND GENERAL PURPOSE CABLE (1501228) is a shielded
twisted pair cable
connecting an IBM PC Network Baseband Adapter or IBM PC Network Baseband
Adapter/A to a non-modular telephone socket. It has the modular connector on
one end while the other end is made for the older screw terminal blocks.
THE IBM CABLING SYSTEM PC NETWORK BASEBAND CABLE (150227) is a shielded
twisted pair cable connecting the IBM PC Network Baseband Adapter, IBM PC
Network Baseband Adapter/A, or the IBM 5173 PC Network Baseband Extender (with
a modular telephone connector) to the data connector wall jack or data
connector distribution panel when using the IBM Cabling System.
Each baseband cable is 7.6 m (25 ft) in length. Several of these cables can be
joined together to obtain cable lengths greater than 7.6 meters, or users can
make their own cables to the lengths they require. With care to prevent
electromagnetic interference, unshielded twisted pair wire, meeting the IBM
Cabling System Type 3 specifications, may also be applied.
THE IBM PC NETWORK BASEBAND EXTENDER RACK MOUNT KIT (1501226) attaches to the
IBM 5173 PC Network Baseband Extender to allow it to be rack mounted in a
standard 19 inch rack.
5.2.2 PC Network Baseband Interfaces
5.2.2.1.1 Medium Access Control support: The IBM PC Network Baseband Adapter
and IBM PC Network Baseband Adapter/A, combined with the adapter support
software of the IBM LAN Support Program, implement CSMA/CD access method
protocols functional equivalent to the IEEE 802.3. (Refer to "2.2.1 Basic
CSMA/CD Concepts" for details on the protocol, and to "5.6 IBM LAN Support
Program" for additional information on LAN Support Program.)
5.2.2.1.2 PC Network Baseband Bridging: The IBM PC Network Baseband network
can be bridged to either the IBM PC Broadband Network or the IBM Token-Ring
Network by using the IBM PC Network Bridge Program. Thus workstations attached
to the PC Baseband network can access stations attached to either of the other
networks, token-ring connected host gateways etc.. More details on the IBM PC
Network Bridge can be found in "6.6.3 IBM PC Network Bridge Program" .
5.2.2.1.3 IEEE 802.2 LLC and Higher Layer Support: The IEEE 802.2 layer is
provided by the IBM LAN Support Program. This program includes the IEEE 802.2
interface, and optionally provides NETBIOS for higher-layer support. The IEEE
802.2 Interface is also required for Advanced Program-to-Program Communication
(APPC). See section "5.6 IBM LAN Support Program" for more details.
5.3 Ethernet/IEEE 802.3
IBM provides support for devices to attach to these LANs. In October 1990 IBM
announced the Personal System/2 Adapter/A for Ethernet Networks. This adapter
attaches to both Ethernet Version 2 and IEEE 802.3 networks.
5.3.1 Workstations
Support for Ethernet or IEEE 802.3 attachment of IBM PCs and PS2s is provided
in both the OS/2 and PC/DOS environment. If running the OS/2 Extended Edition
operating system, Version 1.2, provides the necessary software to enable the
NETBIOS and IEEE 802.2 logical link control application interfaces.
If running PC/DOS, then LAN Support Program Version 1.2 is a requirement.
Further description of this support can be found in "5.6.5 IBM LAN Support
Program Version 1.2."
Ethernet/IEEE attachment is also supported by the PC-RT and RISC RS/6000
processors in the TCP/IP and AIX environments.
5.3.2 Bridges
The IBM 8209 LAN Bridge, with Ethernet Adapter Feature, provides a mechanism
for bridging Ethernet/IEEE 802.3 LAN segments to token-rings. Further
information on this product is available in "6.7 IBM 8209 LAN Bridge."
5.3.3 Gateways
Direct attachment of IBM gateways to Ethernet/IEEE 802.3 LANs is possible if
the IBM 8232 or IBM 3172 products are being used. However, the full range of
IBM communication gateways can be accessed by Ethernet/IEEE 802.3 workstations
if an 8209 LAN bridge is inserted in the network to bridge the CSMA/CD segment
to a token-ring. These products are discussed in "7.3 Non-SNA Gateways."
5.4 Industrial LAN
5.4.1 The 8232 LAN Channel Station and Industrial LAN
The 8232 may also attach to an industrial LAN. The supported adapters are the
INI MP-500 Interface Adapters (15 under heading "5.4.1 The 8232 LAN Channel
Station and Industrial LAN") implementing MAP 2.1.
---Footnote---
(15) Industrial Networking Inc. (INI) is a joint venture between Ungermann-Bass
and General Electric, providing communications products for the manufacturing
environment, today mainly MAP 2.1 implementations.
--------------
Because these adapters require both LAN adapter slots in the 8232 PC, only one
IEEE 802.4 LAN attachment is supported by an IBM 8232 Model 001, and a maximum
of two IEEE 802.4 LAN attachments or one IEEE 802.4 attachment and two IEEE
802.3 or IEEE 802.4 attachments are supported by and IBM 8232 Model 002.
The software requirements are summarized in Figure 68 under heading "5.4.1
The 8232 LAN Channel Station and Industrial LAN", for both the 8232 LAN
Channel Station and the IBM System/370 host.
*---------------------------------------------------------------------*
] 8232: IBM MAP LAN Channel Station Support Program (5798-FBF) ]
] + IBM PC/DOS Support for Token-Bus Adapter (5590-TBI) ]
]---------------------------------------------------------------------]
] IBM non-SNA Host ] LAN Station ]
]-------------------------------+-------------------------------------]
] ] ]
] IBM MAP 2.1 Protocol - ] software support for ]
] VM Support (5798-FBG) ] INI MP-500 MAP Interface Adapter]
] ] (INI 94X4235) ]
] ] ]
*---------------------------------------------------------------------*
Figure 68. 8232 - Software Requirements for MAP 2.1 Support
5.4.2 Migration to MAP 3.0
Migration from MAP 2.1 to MAP 3.0 implies both software application as well as
hardware changes and consequently will require considerable planning.
To minimize the effort for software migration, MAP 2.1 application programming
interfaces (API) in the current VM support on System/370 mainframes have been
designed to be consistent with the MAP 3.0 specifications.
The MMS API Release 2 and the CASE API Release 1 meet the MAP 3.0 Private
Communications User API specifications where applicable. Use of these APIs
will result in fewer programming changes when migrating to MAP 3.0.
The hardware however will require "significant upgrades", including in most
cases replacement of the MAP components. Moreover, future MAP 3.0 products for
workstations are likely to be developed for Family 2 devices, while current
INI MAP 2.1 adapters support only the Family 1 bus design.
5.4.3 MAP - SNA Gateway
Communication between a MAP environment and an SNA network can be provided
only by a gateway function since both networks cover a full seven-layer
communications architecture.
5.4.4 Series/1 and Industrial LAN
The Series/1 in an industrial LAN environment can act as either a
communication server or an application server.
5.4.4.1.1 Communication Server: The communications server runs on the
Realtime Programming System/Communications Manager (RPS/CM) operating system.
Some of the server functions provided are:
o Availability of the Series/1 as a directory server for the network
o Network management agent support
o Host control - that is control of communications with a host system
o Terminal control - that is control of communications with downstream
terminals
o Attachment possibilities for System/370 processors.
With a System/370 processor attached to the Series/1, files can be transferred
to and from the System/370 using the FTAM (File Transfer, Access and
Management) program. Messaging is also supported using the MMFS
(Manufacturing Message Format Standard) support. This provides a way for
user-written messaging applications to build and decode standard messages.
Both FTAM and MMFS are MAP 2.1 Application Layer services (see Figure 35 under
heading "3.2.1.2.2 Manufacturing Automation Protocol"). MMFS is superseded
in MAP 3.0 by the Manufacturing Messaging Specifications (MMS).
The Network Management Agent provides information about the activities in the
local MAP Communications Server system to the network manager.
Host control allows a host system, such as TSO, IMS, or CICS to communicate
with remote 327x terminals, or applications that simulate 327x terminals on a
MAP network. The MAP network is transparent to the host systems.
Terminal control allows users of 327x terminals to communicate with remote
host systems across a MAP network.
5.4.4.1.2 Application Server: The Application Server uses the Series/1 Event
Driven Executive/Communications Facility (EDX/CF) operating system. The
application server is used to control intelligent devices and provides the
following functions:
o Network Management Agent services
o A subset of directory services
o MMFS support
o FTAM subset.
Figure 69 under heading "5.4.4.1.2 Application Server" shows an industrial
LAN with two Series/1 attachments to the bus using a network interface unit
(not shown in the figure).
+----------------------------------------------------------------------------+
] ]
] ]
] *---------* *-----------* *---------* ]
] ] Host 1 ] ] Host 2 ] ] Host 3 ] ]
] ] ] ] ] ] ] ]
] ] OEM ] ]System/370 ] ] OEM ] ]
] ] ] ]-----------] ] ] ]
] ] ] ]FTAM ] MMFS] ] ] ]
] *---------* *-----+-----* *---------* ]
] ] ] ] ]
] *---------* *---------* *---------* ]
] ]Processor] ]Series/1 ]MCS ]Processor] ]
] *---------* *---------* *---------* ]
] ] ] ] ]
] ------------------------------------------------------ ]
] IEEE 802.4 Token Bus ]
] ---------------------------------------------------- ]
] ] ] ] ]
] *---------* *---------* *---------* ]
] ]Processor] MAS]Series/1 ] ]Processor] ]
] *---------* *---------* *---------* ]
] ] ] ] ] ] ] ]
] *-------* ] ] *--------------* ]
] ]Welding] *---------* ]Assembly Line ] ]
] ]Devices] ] Milling ] *--------------* ]
] *-------* ] Plant ] ]
] ] ] ]
] *---------* ]
] ]
] ]
] MCS = MAP Communication Server ]
] MAS = MAP Application Server ]
] ]
+----------------------------------------------------------------------------+
Figure 69. Industrial LAN - Series/1 Servers
5.5 TCP/IP
Although TCP/IP cannot be considered only as a LAN concept or product, the
acronym will be found many times within this document. There is also a fair
sprinkling of TCP/IP application terminology, for instance Telnet. So for
those readers who are not familiar with it, a very short introduction is
offered here. For a more in depth tutorial on the subject, the reader is
recommended to:
o TCP/IP TUTORIAL AND TECHNICAL OVERVIEW - GG24-3376-01
The history of TCP/IP, or TRANSMISSION CONTROL PROTOCOL/INTERNET PROTOCOL can
be traced back to the 1970's, when the United States Defense Advanced Research
Projects Agency (DARPA) funded research into the development of a protocol
suite that would allow networking between computers. The agency, using its
experimental ARPANET network, introduced the concept of a LAYERED architecture
long before the International Standards Organization took any interest in
computer networking. They developed, in conjunction with universities and
other research organizations, the protocols that have become TCP/IP as is
known today. In addition, they funded the implementation of TCP/IP under the
UNIX(**) operating system, and the University of California at Berkeley to
distribute the code free with the operating system. TCP/IP has rapidly spread
among universities and research centers, and has become the standard
communication subsystem for all UNIX connectivity. The UNIX operating system
is implemented by IBM as AIX. In 1983, DARPA demanded that all computers
wishing to connect to the ARPANET should use TCP/IP. This means that TCP/IP is
well supported by the various computer manufacturers, and IBM is no exception.
5.5.1 Structure
As its name implies, TCP/IP consists of two parts:
o The Transmission Control Protocol
o And the Internet Protocol.
TCP/IP also specifies a number of "standard" application functions and
protocols, that can be used for file transfer, simple messaging and terminal
login. These, together with the layering are shown in Figure 70 under heading
"5.5.1 Structure".
-------------------------------------------------------------------------------
*-----------------* *-----------------------------------* *----*
] Applications ] ]SMTP]TELNET]FTP]X-Windows]Rexec]NFS] ]SNMP]
]-----------------] *-------------------------------------------]
] Transport ] ] T C P ] UDP ]
]-----------------] ]-------------------------------------------]
] Internetwork ] ] I P ]
]-----------------] ]-------------------------------------------]
]Network Interface] ] ]
] and ] ]IEEE 802.2-X.25-Satellite-Radio-Asynch-SNA ]
] Hardware ] ] ]
*-----------------* *-------------------------------------------*
-------------------------------------------------------------------------------
Figure 70. TCP/IP Architectural Model
o APPLICATION
Is a user process cooperating with another process on the same or a
different host. Examples include SMTP (Simple Mail Transfer Protocol) and
FTP (File Transfer Protocol).
o TRANSPORT
Provides for the end-to-end data transfer between applications. TCP is a
connection-oriented protocol allowing reliable, acknowledged and sequenced
interchange of information. There is also a connection-less variant, known
as the User Datagram Protocol (UDP). UDP provides a less reliable service
for simpler applications, one of the most important being the Simple
Network Management Protocol (SNMP).
o INTERNETWORK
Provides the networking image, and shields the higher layers from the
physical network below it. The IP deals with the routing of the data
through the network, or when they are interconnected, networks. It does
not provide for reliability of the underlying network, flow control or
error recovery. The IP layer is connection-less.
o NETWORK INTERFACE
is the interface to the actual hardware itself. TCP/IP supports a variety
of physical network types, including IEEE 802.2 Logical Link Control,
X.25, Packet Radio networks, and even SNA. Nothing is assumed about the
reliability of the network to provide reliable delivery of data.
5.5.2 Addressing and Routing
Computers attached to a TCP/IP network, no matter from what the network is
physically built, access each other by an INTERNET ADDRESS As the name
suggests, this function is handled by the IP layer. The address, which may be
considered as a higher layer address has two parts:
o A network address
o And a host address.
Each host computer has a unique address within the network to which it is
attached. If networks are themselves connected together, then each network
must have a unique address.
When networks are connected together, routing devices must be present at the
points of connection to forward traffic from one network to the other. These
special computers, Internet Routers, understand the topology of the Internet
network addressing scheme, and so can forward the data frames towards the
destination network. Applications must know to which network/host addresses
they wish to communicate, or must find out by first interrogating a NAME
SERVER which will translate a user name, for example "host01" to an Internet
address.
However, these Internet addresses do not necessarily bear any relationship to
the addresses which are understood by layers lower than the IP layer, that is
within the network interface. IP supports many different network building
blocks, LANs, X.25 etc., each of which has its own addressing structure. Once
the message has been routed to the correct network, using the IP address, the
last router must be able to map the host address portion of the Internet
address to the lower layer address that will be recognized. This could be a
LAN MAC address, or an X.121 (if using X.25) address. For instance, if a
station is attached to a token-ring, a circulating frame is only copied when
the destination MAC address in the frame is recognized. The station does not
analyze the frame, so cannot recognize Internet addresses. So if the MAC
address is not known to the router, a resolution function must be performed.
This is the function of the ADDRESS RESOLUTION PROTOCOL (ARP).
For ARP, when an Internet address needs to be translated into a physical
address, a special broadcast frame is sent out over the network. Any station
that recognizes its own Internet address within the frame will reply
indicating its physical address. The information is kept by the router in its
ARP CACHE Although the original frame gets discarded, future frames arriving
for that host will be sent to the correct physical destination, in this case,
the token-ring MAC address of the TCP/IP station.
REVERSE ADDRESS RESOLUTION PROTOCOL or RARP works in reverse, much as its name
suggests. Perhaps a device, such as a diskless workstation, does not know its
Internet address when IPL'ed. It may send a RARP request to a server,
indicating its physical address, and enquiring as to what its Internet address
should be set to.
5.5.3 TCP/IP Applications
TCP is a peer-to-peer, connection oriented protocol. There are no master/slave
relations. The applications, however, usually use a server/client model for
communications.
A SERVER is an application that offers a service to users, a CLIENT is the
requestor of that service. An application consists of both a server and a
client part, which can run on the same or on different systems. Users usually
invoke the client part of the application, which builds a request for a
particular service and sends it to the server part of the application, using
the TCP/IP as a transport vehicle.
The server receives the request, performs the service, and replies with the
results to the client. Servers usually can deal with multiple requests at one
time.
o TELNET
The Telnet protocol provides a standardized interface, through which a
program on one host (the Telnet client) may access the resources of
another (the Telnet server) as though the client were a local terminal
attached to the server. The protocol usually works in a line-by-line mode,
similar to that used by many asynchronous terminals. IBM, in its
implementations of Telnet, has introduced working using 3270 datastream,
for obvious reasons. This is known as Telnet/3270.
o FILE TRANSFER PROTOCOL
Copying of files from one machine to another is one of the most frequently
used operations in networking.
The File Transfer Protocol provides the capability to access files and
directories on a remote server host. Files may be transferred in either
direction between server and client.
o X-WINDOWS
The objective of the X-window protocol is to allow applications to execute
across several remote systems and to provide the user with a view into
each system from a single location.
o SMTP
An electronic mail protocol that is used for sending and receiving
messages across the network.
o REXEC/RPC
These protocols have been developed to allow users and applications on one
system to invoke procedure and application execution on remote systems in
the network.
o SNMP
The Simple Network Management protocol is used to communicate management
information between the network management stations and the agents in the
network elements. In many ways, the concepts used here are very similar to
those that have already been discussed in "2.4 LAN Station Management."
The Internet community are investigating moving this protocol in the
direction of CMIP in accordance with the ISO approach.
TCP/IP is supported by IBM in the following environments:
o System/370 Virtual Machine (VM) operating system
o System/370 Multiple Virtual Storage (MVS) operating system
o PC/DOS
o OS/2 Extended Edition
o AS/400 processor
o System/88 processor
o AIX for VM S/370
o AIX for the PC-RT(6150) and RS/6000 processors.
5.6 IBM LAN Support Program
Generally speaking, LAN applications may be developed at different levels
within the communications architecture, either through standard interfaces or
through proprietary communications protocol interfaces.
Some interfaces, although conforming to published standards are system
interfaces accessible only by a low-level programming language such as IBM
PC/DOS Macro Assembler, while other interfaces are described as high-level
Application Programming Language Interfaces (APIs) supported in a high-level
programming language such as C, COBOL, or Pascal. Use of system interfaces is
often limited to systems programming development, where high-level APIs are
more appropriate for general purpose application implementation. Selection of
interfaces to use must take into consideration the degree of product
sensitivity of the API and the susceptibility of the resulting development to
subsequent changes in the supporting environment. In general, users are
encouraged to use higher-level APIs such as those provided by the emerging
System Application Architecture.
In a PC/DOS environment, the IBM LAN Support Program V1.10 provides system
level interfaces interfaces for all current IBM LAN adapters operating in
Family 1 or Family 2 devices. In an OS/2 EE environment, the equivalent
support for Family 2 device adapters only, is provided by the OS/2 EE
Communications Manager.
o IBM Token-Ring Network Adapter (withdrawn from marketing)
o IBM Token-Ring Network Adapter II
o IBM Token-Ring Network Adapter /A
o IBM 16/4 Token-Ring Network Trace and Performance Adapter
o IBM 16/4 Token-Ring Network Trace and Performance Adapter /A
o IBM 16/4 Token-Ring Network Adapter
o IBM 16/4 Token-Ring Network Adapter /A
o IBM 16/4 Token-Ring Network Adapter Enhanced
o IBM 16/4 Token-Ring Network Adapter /A Enhanced
o IBM PC Network Adapter (withdrawn from marketing)
o IBM PC Network Adapter II
o IBM PC Network Adapter II/A
o IBM PC Network Adapter II Frequency 2
o IBM PC Network Adapter II/A Frequency 2
o IBM PC Network Adapter II Frequency 3
o IBM PC Network Adapter II/A Frequency 3
o IBM PC Network Baseband Adapter
o IBM PC Network Baseband Adapter /A.
The IBM LAN Support Program implements IEEE 802.2 logical link control layer
services and protocols, supporting the standard logical link control Class II
open interface. Optionally, IBM LAN Support Program provides a NetBIOS
higher-layer interface for the IBM Token-Ring Network, IBM PC Network
(Broadband) and IBM PC Network Baseband, and with the announcement of IBM LAN
Support Program Version 1.2 provides these interfaces to Ethernet Version 2 or
IEEE 802.3 networks as well. Support for these two networks was introduced by
OS/2 Extended Edition in release 1.2 of that product. The term ETHERAND(*) is
used by IBM to describe this function in the OS/2 environment. Further
information can be found on LAN Support Program and Ethernet support in "5.6.5
IBM LAN Support Program Version 1.2." Application programs can be written
against the IEEE 802.2 or NetBIOS interfaces. In addition, a number of other
higher-layer protocols and interfaces may operate on top of the open LLC
interface, including IBM's strategic LU 6.2 protocol for program-to-program
communication. Such additional higher layer protocols may be provided through
separate products which use IBM LAN Support Program, for example, APPC/PC for
LU 6.2 communications.
NetBIOS is a proprietary but widely implemented higher-level protocol LAN
interface for workstation applications. Various applications are available
which use the NetBIOS interface to communicate across the LAN, for example the
IBM PC LAN Program 1.3 and IBM PC 3270 Emulation Program Version 3 (Gateway
and Network Station configurations).
All the function supplied by the IBM LAN Support Program V1.10 is included in
the OS/2 Extended Edition 1.1 (trademark of the International Business
Machines Corporation) Communications Manager. In addition, the LU 6.2
implementation included in OS/2 Extended Edition Communications Manager has a
much more convenient programming language interface as well as high-level
language support (IBM C/2 Compiler and IBM Pascal/2 Compiler). The programming
interface included in the PC/DOS APPC/PC is limited to IBM PC/DOS Macro
Assembler.
Figure 71 under heading "5.6 IBM LAN Support Program" shows the base
programming interfaces on IBM LANs as introduced above.
+----------------------------------------------------------------------------+
] ]
] ]
] Interfaces Protocols SNA Layers ]
] ]
] *----------------------* ]
] ] Transaction ] ]
] LU 6.2 ------------------>*--------* ]----------------------] ]
] ] ] ] Presentation ] ]
] NetBIOS -------->*-------*] ] ]----------------------] ]
] ] ]] LU 6.2 ] ] Data Flow Control ] ]
] ] ]] ] ]--- ---] ]
] ]NetBIOS]] ] ] Transmission Control ] ]
] ] ]]--------] ]--- ---] ]
] ] ]] PU 2.1 ] ] Path Control ] ]
] 802.2 ---->*-----------------------] ]----------------------] ]
] ] IEEE 802.2 LLC ] ] Data Link ] ]
] MAC -->*---------------------------] ] Control ] ]
] ] medium access control ] ]----------------------] ]
] *---------------------------* ] Physical ] ]
] *----------------------* ]
] ]
] ---------------------------------------------------------* ]
] OS/2 EE 1.1 ] Communications Manager ] ]
] ------------+--------------------------------------------] ]
] PC/DOS ] APPC/PC ] LAN Support Program ] ]
] ------------+---------+----------------------------------] ]
] Interfaces ] LU 6.2 ] NetBIOS ] 802.2 LLC ] Direct MAC ] ]
] ---------------------------------------------------------* ]
] ]
] ]
+----------------------------------------------------------------------------+
Figure 71. IBM LANs - Base Programming Interfaces
The adapter support program, as provided by the IBM LAN Support Program V1.10
or as included in the LAN Communications Manager of OS/2 Extended Edition 1.1
or higher, provides a system application programming interface to the adapter.
With the use of an adapter support program, a local area network system
application program assembles a control block containing a command and related
information for the adapter. Control is passed to the adapter support program
and the application awaits the results. In other words, the application
program passes command control blocks (CCBs) to the adapter support program to
interface with the adapter.
Command control blocks are used at the level of the IEEE 802.2 interface and
the direct (medium access control) interface.
With the advent of the OS/2 Extended Edition 1.1 LAN Communications Manager,
two additional, slightly different command control block structures have been
added to the original CCB as it existed in the IBM LAN Support Program. One
new CCB is associated with OS/2 1.1's dynamic link routine interface, the
other one with its device driver interface.
Applications coded at the MAC level (through the direct interface) are
confined to the local ring and will not be able to communicate across bridges.
Additional APIs are offered with various LAN applications, for example, the PC
3270 Emulation API, the Server/Requester Programming Interface (SRPI),the 3270
Workstation Program HLLAPI or OS/2's Common Services Interface (CSI).
5.6.1 Base Programming Interfaces
5.6.1.1 Open IEEE 802.2 LLC Interface
An overview of the IEEE 802.2 logical link control sublayer was described in
"3.2.2 Logical Link Control Sublayer," including the concept of LLC service
access points to interface between data link control and higher layer protocol
services.
Both connectionless or datagram services as well as connection-oriented
services are included in the standard. The IBM LAN Support Program supports
both, as Class II operation. The format of the LLC protocol data unit (LPDU)
and a list of the service commands/responses is included in "3.2.2.3 LLC
Protocol Data Unit."
5.6.1.2 NetBIOS Interface
NETWORK BASIC INPUT OUTPUT SYSTEM (NetBIOS) is a well-accepted proprietary
higher-level LAN protocol, providing a session-level programming interface.
NetBIOS supports a layered communications architecture as shown in Figure 72
under heading "5.6.1.2 NetBIOS Interface". Names rather than network
addresses are used by Netbios applications for session support. NetBIOS
support provides services to locate and associate these names with MAC network
addresses.
+----------------------------------------------------------------------------+
] ]
] ]
] ]
] LAN station A LAN station B ]
] ]
] ] NetBIOS ] ] NetBIOS ] ]
] ] Interface ] ] Interface ] ]
] *--------------* *--------------* ]
] ] Session ]<-----Message---->] Session ] ]
] ]--------------] ]--------------] ]
] ] Transport ]<---Connection--->] Transport ] ]
] ]--------------] ]--------------] ]
] ] Network ]<-----Packet----->] Network ] ]
] ]--------------] ]--------------] ]
] ] Link ]<----- LPDU ----->] Link ] ]
] ]--------------] ]--------------] ]
] ] Physical ]<---------------->] Physical ] ]
] *--------------* LAN Physical *--------------* ]
] Medium ]
] ]
] LPDU = LLC protocol data unit ]
] ]
+----------------------------------------------------------------------------+
Figure 72. NetBIOS Layer Support
o LINK LAYER
At the Link layer, NetBIOS can use "connectionless" (user datagram) or
connection-oriented services. This layer is responsible for the assembly
of data units, and the delivery of these data units to the physical media
for transmission. CRC checking is also performed.
o NETWORK LAYER
The network layer is implemented as a "null" layer.
o TRANSPORT LAYER
At this level, point-to-point connections are created supporting data
transmission and acknowledgements. Any required flow control or pacing is
handled at this level.
o SESSION LAYER
Through the session layer, NetBIOS establishes "name" pairs. A name is an
identifier for a logical entity in which all session-level communication
activity is centered.
The NETBIOS INTERFACE provides an application programming interface offering
commands for session control, name support, datagram support, and debugging
facilities. An overview of the NetBIOS interface and functions is provided
later of this section. For additional detail and programming support, users
should refer to the IBM LOCAL AREA NETWORK TECHNICAL REFERENCE.
5.6.1.2.1.1 NetBIOS Implementations: Two different versions of NetBIOS,
including different error codes and transmission sizes, currently exist and
are incompatible from the standpoint of interoperability. NetBIOS was
originally provided in microcode on the IBM PC Network Adapter in Read Only
Memory (ROM). Subsequently a superset of the original NetBIOS was implemented
in the IBM LAN Support Program V1.10, providing equivalent function and
application interfaces for all IBM LAN-attached workstations under control of
PC/DOS.
When, on a IBM PC Network (Broadband), some stations are equipped with the IBM
PC Network Adapter while others are equipped with second generation PC Network
adapters, a NetBIOS coexistence problem arises which must be solved to allow
communication among all PC Network (Broadband) stations. Two solutions may be
considered, each with different consequences. They will be presented in
section "5.6.2 LAN Support Program Structure."
Figure 71 under heading "5.6 IBM LAN Support Program" shows the NetBIOS
module as it is implemented by the IBM LAN Support Program V1.10. This module
interfaces to IEEE 802.2 via the X'F0' LLC service access point (see "3.2.2.2
IEEE 802.2, ISO 8802-2"). NetBIOS provides services supporting (non-SNA)
peer-to-peer communications.
5.6.1.2.1.2 The NetBIOS Interface: The NetBIOS function within each
workstation maintains a table of names that this node and its session partners
are known by on the network. Communications between NetBIOS applications is
based upon resource names. These names are provided to NetBIOS by the
application program. A name can be a unique or a group name. NetBIOS provides
services to ensure uniqueness of names. If name is in the NetBIOS Name Table,
a session can be established from this named resource with another named
resource. NetBIOS maintains up to 254 selectable names plus one permanent node
name (10 bytes of binary zeros followed by the adapter's burned-in address).
NetBIOS is operated from an application program through a control block,
called the NETWORK CONTROL BLOCK (NCB).
The NetBIOS Interface provides the following services:
GENERAL CONTROL SERVICES. which include:
o RESET
In the LAN Support Program implementation, this command aborts all
sessions, clears both the session and name tables and resets the NetBIOS
status. In OS/2 Extended Edition 1.1, this command aborts all sessions
for that application program process, clears both the session and name
tables and resets the NetBIOS status.
o STATUS
This command requests the status of either a local or remote NetBIOS
resource for transfer to the user's program area.
o CANCEL
This command requests that a command associated with a given network
control block (NCB) be cancelled.
o LAN STATUS ALERT
This command is used by application programs that wish to be notified of
temporary LAN error conditions that last for over one minute.
o UNLINK
This command is provided only for NetBIOS compatibility. In the PC
Network it drops a session for Remote Program Load. Where the command is
issued to an adapter that does not support Remote Program Load, the
command will be treated by the NetBIOS program as a "no-operation".
NAME SUPPORT SERVICES. which allow a program to manage user-assigned names or
identifiers by use of the following commands:
o ADD NAME
To add a new name (up to 16 characters) to the memory resident name table.
o ADD GROUP NAME
A "Group Name" is a mechanism for allowing more than one station to share
the same resource name. This enables resources to be defined for
functions that may exist in multiple workstations in a distributed
application environment.
o DELETE NAME
To remove a name from the name table.
o FIND NAME
To determine whether a given name is registered on a network adapter.
SESSION SUPPORT SERVICES. A "session" is a logical connection between two
named resources supporting peer-to-peer communications. When a session has
been established with another named resource, information can be exchanged
over the session between the two resources. This reliable data transfer is
provided by the SESSION layer. A program may refer to multiple names and
therefore sustain multiple sessions between itself and other stations on the
LAN.
Command services available to the user to manage the sessions include:
o CALL
This command opens a session with another named resource which has an
outstanding Listen command.
o LISTEN
This command enables a session to be established between the named
resource that issues the Listen and any named resource on the LAN which
issues a Call.
o HANG UP
This command closes a session between two named resources.
o SEND
This command is used to transmit data between two session partners. The
maximum MESSAGE size is 64 Kbytes. To maintain data integrity, the
session is closed if a Send command fails.
o CHAIN SEND
A Chain Send is equivalent to the Send command, but the message is made up
of two concatenated buffers (maximum 64 Kbytes in total).
o SEND NO ACKNOWLEDGEMENT Like the Send command this command provides a send
facility, but without requiring a NetBIOS level data acknowledgement.
o CHAIN SEND NO ACKNOWLEDGEMENT
This command is equivalent to the Send No Acknowledgement command, but the
message is made up of two concatenated buffers (maximum 64 Kbytes in
total).
o RECEIVE
This command receives data sent over a specified session or over any open
session.
o RECEIVE ANY
This command receives data sent from any session partner.
o SESSION STATUS
This command will return the status of one or all sessions for a given
Name.
DATAGRAM SUPPORT SERVICES provide a connectionless data transmission service.
That is, when individual unrelated messages (called datagrams) are to be sent,
it is not always necessary to establish a session between the network
stations. Messages are just transmitted to a named station or a group of
stations on the LAN. No error recovery or sequencing is performed and a
station does not have to acknowledge receiving a frame. Datagram Support
Services can be contrasted with the "connection-oriented" services offered by
the Session Support Services commands. Data transfer using datagram support
goes directly to the DATA LINK layer.
o SEND DATAGRAM
Allows an application program to send a datagram to a specific name or
group name.
o SEND BROADCAST DATAGRAM
Allows an application program to send a datagram to any station which has
a Receive Broadcast Datagram command outstanding.
o RECEIVE DATAGRAM
A station must use this command to receive datagrams from any name on the
network that issues a Send Datagram.
o RECEIVE BROADCAST DATAGRAM
A station must use this command to receive datagrams from any name on the
network that issues a Send Broadcast Datagram.
DEBUGGING SUPPORT SERVICES
o TRACE
In the IBM LAN Support Program V1.10 and OS/2 EE 1.1 implementations of
NetBIOS this command has been added to activate and de-activate a trace of
all the network control blocks (NCBs) issued to NetBIOS by the application
program and some of the command control blocks (CCBs) issued by NetBIOS.
5.6.1.3 LU 6.2 Interface
APPC (Advanced Program to Program Communication) defines a set of
communication services that enable transaction programs to cooperatively
execute distributed transactions. APPC function is implemented in an SNA
protocol called LU Type 6.2. APPC Transaction Programs (TP), that is
application programs using the APPC interface, use the LU 6.2 protocols to
communicate with each other as peers.
Communication between two APPC programs is known as a "conversation". A
conversation flows on an underlying session between two LUs. When parallel
sessions are supported between the LUs, multiple conversations can run
concurrently between two transaction programs. Parallel sessions are only
supported for LUs associated with Type 2.1 node devices.
Products that implement LU 6.2 provide APPC services in different ways. The
precise definition of these services is independent of any particular product,
and constitutes part of the architectural definition of LU 6.2. The services
are invoked by defined verbs, parameters, and return codes. A "verb"
corresponds to a request by a transaction program for some specific service
from the LU. As long as a transaction only requests APPC functions supported
by both systems (peers), a programmer can use the APPC verb syntax to develop
a distributed application independent of the system it will run on.
There are two categories of LU 6.2 conversation verbs, BASIC and MAPPED.
Basic conversation verbs provide a lower-level interface to the LU. These
verbs have greater privileges and are appropriate for system-oriented
transaction programs that perform services for other transaction programs.
Mapped conversation verbs are intended for user-written transaction programs.
They provide an interface suitable for high level languages. Conversations
are called basic or mapped conversations depending on the verbs being used.
APPC is IBM's strategic architecture for distributed transaction processing,
and has been implemented in the following products. These products all
support APPC mapped conversations:
o CICS/VS
o AS/400
o S/38
o System/36
o Series/1
o PCs, PS/2s (APPC/PC, OS/2 EE).
The System Application Architecture (SAA) Common Programming Interface for
Communications is a published application programming interface that invokes
the services of APPC and LU 6.2 from high-level languages in various SAA
environments. As SAA implementations evolve, it is intended that the defined
programming interfaces enable users to develop SAA conforming applications
that can exist in different SAA product environments with greater product
transparency to both programmers and users. Although the Common Programming
Interface for Communications is not yet implemented in all the target
environments, users can develop APPC applications consistent with the
published interface to minimize ongoing application changes.
5.6.1.3.1 APPC/PC: APPC/PC, available in a PC/DOS environment, provides Type
2.1 node and/or PU 2.0 and LU 6.2 function to allow transaction programs to be
written for IBM workstations to communicate with other APPC systems. These
systems could be any of those listed above; the communication could take place
over a LAN or an SDLC link.
LAN workstations attached to the IBM PC Network (Baseband and Broadband) or
IBM Token-Ring Network can communicate with each other, with an AS/400,
System/36 (5363) or System/370 directly attached to the LAN. If the LAN is
remote to the host system, or is attached to the LAN using a gateway to
communicate, the APPC transaction program must reside in the gateway, or a
relay program must be written to run in the gateway.
This relay program is needed because APPC/PC does not have the ability to
examine an incoming connection request and re-route it to another APPC
application (see Figure 73 under heading "5.6.1.3.1 APPC/PC").
+------------------------------------------------------------------------------
]
]
] *---*
] ] ]APPC/PC
] *---*
] :]
] APPC/PC *------:----------* System/370 host
] *---*.........: .........*---* *--------*
] ] ]--] : ]--] ]------------ ] ]
] *---* ] : ] *---*.........../........] CICS ]
] *-----------:-----* APPC ---------] ]
] ]: relay ] ]
] *---* program *--------*
] ] ] CICS APPC
] *---* application
] APPC
] transaction
] program
]
]
+------------------------------------------------------------------------------
Figure 73. APPC/PC Connections
Through its application programming interface, APPC/PC provides support for
both mapped and basic conversations.
Parallel sessions are supported when communicating with the following systems:
o AS/400
o System/36
o S/38
o Series/1
o Another APPC/PC.
5.6.1.3.1.1 Software Requirements: APPC/PC requires the IBM LAN Support
Program V1.10 to provide the IEEE 802.2 LLC protocol and interface.
5.6.1.3.2 OS/2 Extended Edition LU 6.2 Support: APPC support of the OS/2 EE
Communications Manager includes not only the base functions of the LU 6.2
protocol, but also provides additional features.
In this environment, a user does not have to issue control verbs like
Attach_PU or Attach_LU, as is required when using APPC/PC under PC/DOS. The
Application Subsystem shipped with Communications Manager uses the
configuration file to perform the establishment of the SNA node and all the
links with the other nodes. It also manages activation and deactivation of
conversations by means of the automatic start of transaction programs (TPs).
The major features of APPC support under OS/2 Extended Edition are:
o Connectivity via SDLC and IBM LAN adapters
o Acting as either a peer-to-peer (Type 2.1) or boundary function node (PU
2.0)
o Link sharing with 3270 emulation LUs (LU 2.0)
o Support of multiple LUs
o Implicit partner LU support
o Support of parallel sessions
o Automatic session activation
o Remotely initiated TPs
o Multiple access by different TPs
o Use of locally known names (for example, LU aliases)
o Communication and System Management (C&SM) support
o Sending of alerts to the network resource specified in the configuration
file
o Link congestion control
o Programming languages support:
- Macro Assembler
- Pascal
- C language.
5.6.2 LAN Support Program Structure
LAN Support Program modules are called DXMnnMOD.SYS where (nn =
A0,C0,C1,T0,etc), which will be installed as device drivers (that is,
extensions of the PC/DOS environment) under IBM PC/DOS 3.3 or 4.0.
If two LAN adapters are installed in one LAN station, only one copy of each
required device driver needs to be loaded.
The A0 module, also called the interrupt arbitrator, is common for any IBM LAN
adapter. It provides the appropriate adapter support interface.
The A0 module allows one optional parameter, the language for load-time error
messages:
o 01 - US English (default)
o 44 - UK English
o 33 - French
o 49 - German
o 39 - Italian
o 34 - Spanish
o 46 - Swedish.
The LAN Support Program also contains:
o DXMAID.EXE: the LAN Support Program Installation Aid.
o DXMINFO.DOC: the LAN Support Program Online Documentation.
o TIMERINT.SYS: a device driver to replace obsolete ROM timer interrupt code
on old PCs (a BIOSDATE.EXE is provided to determine the requirement for
this driver).
The installation aid will place the required device driver commands and
default parameters in the CONFIG.SYS file according to the type of LAN adapter
and the answers supplied while running the Installation Aid.
5.6.2.1.1 IBM Token-Ring Network and LAN Support Program: The IBM LAN
Support Program V1.10 supports the new 16/4 Mbps IBM Token-Ring Network
adapters.
Support for the 64Kbytes of RAM, including implementation of RAM paging is
provided. RAM Paging enables use of the entire 64 Kbytes shared RAM on the
adapter while only requiring 16 Kbytes in the PC memory. This limits the risk
of address range conflicts with other installed adapters, for example, a
second LAN adapter.
Both the data rate and the required shared RAM size are specified at adapter
installation time via hardware switch settings on the Family 1 adapter, or a
software adapter configuration file update for a Family 2 adapter.
The structure of the IBM LAN Support Program V1.10 modules when supporting IBM
Token-Ring Network adapters for token-ring attached workstations is shown in
Figure 74 under heading "5.6.2.1.1 IBM Token-Ring Network and LAN Support
Program":
+------------------------------------------------------------------------------
]
]
] * - - - > *---* *---* <-- NetBIOS Interface
] ] ]T0 ] ]T0 ]
] ] *---* *---*
] LAN *-----------* *-----------* <-- 802.2 LLC Interface
] Support ] C0 ] ] C1 ]
] Program *-----------* *-----------*
] ] *-------------------------* <-- Direct Interface
] ] ] A0 ]
] * - - - > *-------------------------*
] * - - - > *-------**-------**-------*
] ] ] TRN ]] TRN ]]16/4 TR]
] Adapters ]Adapter]]Adapter]]Adapter]
] ] ] ]]II , /A]]II , /A]
] * - - - > *-------**-------**-------*
] <--- Token Ring Network -->
]
]
+------------------------------------------------------------------------------
Figure 74. LAN Support Program and IBM Token-Ring Network
IBM LAN Support Program will provide the following device driver commands:
DEVICE=\path\DXMA0MOD.SYS
DEVICE=-path-DXMCnMOD.SYS
local-addr-0,RAM-0,etr-0,local-addr-1,RAM-1,etr_1
DEVICE=\path\DXMT0MOD.SYS p1=n1 p2=n2 ... / q1=m1 q2=m2 ...
The C0 module supports IEEE 802.2 Class II operation and provides the
appropriate interface.
The C1 module supports IEEE 802.2 Class II operation and provides the
appropriate interface for token-ring attached 3270-PCs and LAN stations
running 3270 Workstation Program.
When operating at 16 Mbps, use of the early token release option of the token
passing ring MAC protocol is indicated by a parameter for the DXMCnMOD.SYS
module in the CONFIG.SYS data set.
Both Cn modules support positional parameters to specify operational values
for the primary adapter (0) and an alternate adapter (1) if present:
o Local addr: locally administered address.
o RAM: mapping address for shared RAM (ignored for Family 2 Token-Ring
adapters).
o ETR: early token release, 0 to use (default) or 1 to disable early token
release (ignored if operating at 4 Mbps).
The T0 module, which provides the common NetBIOS protocol and interface uses
the LLC Interface provided by the Cn module through a specific NetBIOS service
access point (SAP X'F0').
The NetBIOS module parameters will not be discussed in detail in this
document, as they are related to installation parameters of specific NetBIOS
applications such as the IBM PC LAN Program and the IBM PC 3270 Emulation
Program (Gateway and Network Station configurations). Refer to DXMINFO.DOC,
the LAN Support Program Online Documentation file, and to the Raleigh ITSC
publication INSTALLATION GUIDELINES FOR IBM TOKEN-RING NETWORK PRODUCTS.
5.6.2.1.2 IBM PC Network (Broadband, Baseband) and LAN Support Program: The
structure of the IBM LAN Support Program V1.10 modules when supporting IBM PC
Network (Broadband) or IBM PC Network Baseband is shown in Figure 75 under
heading "5.6.2.1.2 IBM PC Network (Broadband, Baseband) and LAN Support
Program".
+------------------------------------------------------------------------------
]
]
]* - - - > *--**---* *---* *---* *---* *---* <-- NetBIOS I/F
]] ] ]]T0 ] ]T0 ] ]T0 ] ]T0 ] ]T0 ]
]] ] ]*---* *---* *---* *---* *---*
]LAN ] ]*---**-------**-------**-------**-------* <-- 802.2 I/F
]Support ] ]]G2 ]] G1 ]] G0 ]] G1 ]] G0 ]
]Program ] ]*---**-------**-------**-------**-------*
]] ] ]*---------------------**----------------* <-- Direct I/F
]] ] ]] A0 ]] A0 ]
]* - - - > ] ]*---------------------**----------------*
]* - - - > ] *----**-------**-------**-------**-------*
] ]PC Net ]]PC Net ]]PC Net ]]PC Net ]]PC Net ]
]Adapters ](orig.)]]II Freq]]II/A Fr]]Baseb. ]]Baseb. ]
] ] ]]1, 2, 3]]1, 2, 3]] ]] /A ]
]* - - - > *-------**-------**-------**-------**-------*
] <-----PCNet Broadband-----><-PCNet Baseband->
]
]
+------------------------------------------------------------------------------
Figure 75. LAN Support Program and IBM PC Network
IBM LAN Support Program will provide the following device driver commands:
DEVICE=\path\DXMA0MOD.SYS
DEVICE=-path-DXMGnMOD.SYS local-addr-0,wrk-0,local-addr-1,wrk-1
DEVICE=\path\DXMT0MOD.SYS p1=n1 p2=n2 ... / q1=m1 q2=m2 ...
The Gn modules (n = 0, 1, 2; 31, 34, 38 Kbytes resp.) provide the same IEEE
802.2 LLC interface on PC Network adapters as the Cn modules provide on
token-ring adapters. The G1 module is required to run Workstation Program on
a LAN Station (or to operate the obsolete 3270-PC on a PC Network). The G2
module will bypass the ROM NetBIOS on an original PC Network Adapter.
The Gn PC Network Device Drivers allow a maximum internal work area of 64
Kbytes per adapter. The amount of work space actually required depends on the
requirements of the application program with respect to the number of
concurrent NetBIOS sessions. The default work space is 8 Kbytes for each PC
Network Adapter and is valid for up to fifteen sessions. Between fifteen and
twenty-three sessions, a work space of 12 Kbytes is needed, while 16 Kbytes
will be used if twenty-four to thirty-two sessions are used concurrently. The
WRK parameter of DXMGnMOD.SYS defines the desired internal work area size (in
Kbytes).
A common NetBIOS Interface device driver T0 (24 Kbytes), is accessed by the
LLC Interface of the Gn module through a specific NetBIOS service access point
(SAP X'F0').
5.6.2.1.3 PC Network (Broadband) Coexistence: Since the NetBIOS support
residing in ROM on the original PC Network (Broadband) adapters is
incompatible with the NetBIOS support required for the later PC Network
(Broadband) II and II/A adapters, (module DXMT0MOD.SYS), two alternatives
exist to address requirements for the devices to interoperate on a PC Network
(Broadband).
The DXMG2MOD.SYS module of the IBM LAN Support Program V1.10 bypasses the ROM
NetBIOS on the original PC Network (Broadband) adapter. It enables these
adapters to use the IEEE 802.2 LLC interface to access other higher-level
protocols such as APPC or the enhanced NetBIOS provided by the IBM LAN Support
Program V1.10 This alternative requires processor memory for the additional
LAN Support Program device drivers.
The PC NETWORK PROTOCOL DRIVER program on a PC Network (Broadband) Adapter II
or II/A provides a NetBIOS level which is compatible with the ROM NetBIOS on
an original PC Network Adapter.
This solution does not provide an open LLC interface and degrades the NetBIOS
performance possible with the IBM LAN Support Program V1.10 Therefore it
should only be considered if there are severe memory limitations on the PCs
containing the original PC Network Adapters which would prevent them from
using the IBM LAN Support Program V1.10.
5.6.2.1.4 IBM LAN Support Program Structure -Summary: Whenever a workstation
contains two LAN adapters, only one copy of each common LAN Support Program
module is required.
IBM LAN Support Program requires DOS 3.3 or 4.0. Figure 76 under heading
"5.6.2.1.4 IBM LAN Support Program Structure -Summary" shows how the LU 6.2
interface is made available through the IBM LAN Support Program V1.10 combined
with the APPC/PC program.
+------------------------------------------------------------------------------
]
]
]- - LU 6.2 - - - LU6.2 - - ] ] - - LU 6.2 - - - LU 6.2 - - - LU 6.2 - -
]*-------------------------**--**---------------------**----------------*
]] APPC/PC ]] ]] APPC/PC ]] APPC/PC ]
]*-------------------------*] ]*---------------------**----------------*
] ] ]
]- 802.2 LLC - - 802.2 LLC -] ]- 802.2 LLC - - 802.2 LLC - - 802.2 LLC -
]*-----------* *-----------*] ]*---**-------**-------**-------**-------*
]] C1 ] ] C0 ]] ]]G2 ]] G1 ]] G0 ]] G1 ]] G0 ]
]*-----------* *-----------*] ]*---**-------**-------**-------**-------*
]*-------------------------*] ]*---------------------**----------------*
]] A0 ]] ]] A0 ]] A0 ]
]*-------------------------*] ]*---------------------**----------------*
] ] ]
]*-------**-------**-------*] *----**-------**-------**-------**-------*
]]T-R Net]]T-R Net]]16/4 TR]]PC Net ]]PC Net ]]PC Net ]]PC Net ]]PC Net ]
]]Adapter]]Adapter]]Adapter]](orig.)]]II Freq]]II/A Fr]]Baseb. ]]Baseb. ]
]] ]]II , /A]]II , /A]] ]]1, 2, 3]]1, 2, 3]] ]] /A ]
]*-------**-------**-------**-------**-------**-------**-------**-------*
]<-------Token-Ring--------><-----PCNet Broadband-----><-PCNet Baseband->
]
]
+------------------------------------------------------------------------------
Figure 76. LAN Support Program and APPC/PC
APPC/PC based applications may be run on any IBM LAN. LU 6.2 support is
accessed by the appropriate IBM LAN Support Program V1.10 LLC module through
an SNA service access point (X'04', X'08', ...).
APPC/PC is a separate product and can be also be run over an SDLC link.
5.6.3 OS/2 Extended Edition 1.1 LAN Support Structure
Figure 77 under heading "5.6.3 OS/2 Extended Edition 1.1 LAN Support
Structure" shows the Communications Manager interfaces, including those for PC
Network (Broadband), PC Network Baseband and the IBM Token-Ring Network.
These LAN interfaces are very similar to the interfaces provided by IBM LAN
Support Program V1.10. Most of the commands are identical and the structure
of the commands is very similar.
As mentioned before, specific command control blocks (CCBs) are used to pass
commands to the adapter support for OS/2 Extended Edition Communications
Manager LAN support.
+----------------------------------------------------------------------------+
] ]
] ]
] *----------------------------------------* ]
] ] ] ]
] ] Network Application Programs ] ]
] ] *-------------* ] ]
] ] ] APPC ]SRPI] ] ]
] *-----] ](LU 6.2)]----------* ] ]
] ]3101 ] ] ]3270 Emul.] ] ]
] ]VT100] *-----------+----------------------] ]
] ]Emul.] ] NetBIOS ] Common Service ] ]
] ]-----------------------] (Service Verbs) ] ]
] ] ACDI ] IEEE 802.2 ] ] ]
] ]----------------------------------------------] ]
] ] Operating System/2 Kernel ] ]
] ]----------------------------------------------] ]
] ] Async ] PC Net ] Token-Ring ] SDLC ] DFT ] ]
] *----------------------------------------------* ]
] ] ] ] ] ] ] ] ] ] ] ]
] *----------------------------------------------* ]
] ] Communications Adapter's I/O ] ]
] *----------------------------------------------* ]
] ]
] ]
+----------------------------------------------------------------------------+
Figure 77. OS/2 Extended Edition 1.1 - Communications Manager Interfaces
One operational consideration is that, whereas in the PC/DOS environment the
adapter support and programming interfaces are separate product(s) which must
be loaded after loading the operating system, in OS/2 Extended Edition 1.1 the
adapter support program is an integral part of the operating system.
As in the IBM LAN Support Program V1.10, the NetBIOS implementation in the
Communications Manager of Operating System/2 Extended Edition V1.1 (trademark
of the International Business Machines Corporation) runs on top of the IEEE
802.2 interface. It is used primarily to communicate between PC LAN Program
requesters and servers or between Operating System/2 Extended Edition V1.1
requesters and the OS/2 LAN Server. It is also available to network-specific
applications.
The NetBIOS commands provided by the Communications Manager of Operating
System/2 Extended Edition V1.1 are identical to the ones described earlier for
IBM LAN Support Program V1.10 (see "5.6.1.2 NetBIOS Interface").
In an Operating System/2 Extended Edition V1.1 environment, IBM local area
networks may be accessed indirectly through OS/2 dynamic link routines or
directly via device driver calls. Two sublayer interfaces are available, IEEE
802.2 logical link control and IEEE 802.5 medium access control (on IBM
Token-Ring Network only).
The 802.2 LLC interface supports both connectionless and connection-oriented
services. A multi-thread programming interface provides support for multiple
application users. It handles control blocks passed from NetBIOS, from APPC
and from user applications concurrently.
5.6.3.1.1.1 LAN support installation and configuration: Operating System/2
Extended Edition V1.1 uses configuration panels to define installation
parameters for workstations attached IBM Token-Ring Network adapters and PC
Network adapters. These panels are presented to the user depending upon the
workstation configuration data.
Figure 78 under heading "5.6.3.1.1.1 LAN support installation and
configuration" illustrates the profile panels for a token-ring adapter and
Figure 79 under heading "5.6.3.1.1.1 LAN support installation and
configuration" shows the profile panels for a PC Network adapter.
+----------------------------------------------------------------------------+
] ]
]*----------------------------------------------------------------------* ]
]] LAN IEEE 802.2 Profile for Token-Ring (1 of 2) ] ]
]] ] ]
]] Adapter number and version . . . . . : n - /A ] ]
]] ] ]
]] Adapter shared memory . . . . . . . : ] ]
]] Use Universally administered address : Yes ] ]
]] Adapter address . . . . . . . . . . : ] ]
]] Functional address . . . . . . . . . : ] ]
]] Group address . . . . . . . . . . . : ] ]
]] ] ]
]] Maximum number SAPs . . . . . . . . : 2 ] ]
]] Maximum link stations . . . . . . . : 8 ] ]
]] Maximum number group SAPs . . . . . : 0 ] ]
]] Maximum members per group SAP . . . : 0 ] ]
]] Maximum number of users . . . . . . : 2 ] ]
]] ] ]
]] Transmit buffer size . . . . . . . . : 1048 bytes ] ]
]] Number of transmit buffers . . . . . : 2 ] ]
]] Receive buffer size . . . . . . . . : 280 bytes ] ]
]] Minimum receive buffers . . . . . . : 10 ] ]
]] ] ]
]] System key (hex character) . . . . . : ] ]
]] ] ]
]]----------------------------------------------------------------------] ]
]] Esc=Cancel F1=Help F8=Forward ] ]
]*----------------------------------------------------------------------* ]
]*----------------------------------------------------------------------* ]
]] Pass attention MAC frames default . . . : No ] ]
]] Monitor contention default . . . . . . . : No ] ]
]] Pass beacon frames default . . . . . . . : No ] ]
]] ] ]
]] Group 1 Response Timer . . . . . . . . . . : 50 x 40 ms. ] ]
]] Group 1 Acknowledgement Timer . . . . . . : 50 x 40 ms. ] ]
]] Group 1 Inactivity Timer . . . . . . . . . : 255 x 40 ms. ] ]
]] Group 2 Response Timer . . . . . . . . . . : 50 x 40 ms. ] ]
]] Group 2 Acknowledgement Timer . . . . . . : 50 x 40 ms. ] ]
]] Group 2 Inactivity Timer . . . . . . . . . : 255 x 40 ms. ] ]
]] ] ]
]] Common LAN adapters queue size . . . . . . : 400 elements ] ]
]] Number Global Descriptor Table Selectors . : 50 ] ]
]] Attended mode of initialization . . . . . : No ] ]
]] ] ]
]]----------------------------------------------------------------------] ]
]] Esc=Cancel F1=Help F7=Backward ] ]
]*----------------------------------------------------------------------* ]
+----------------------------------------------------------------------------+
Figure 78. OS/2 EE 1.1 - IEEE 802.2 Profile for Token-Ring
+----------------------------------------------------------------------------+
] ]
] ]
]*----------------------------------------------------------------------* ]
]] LAN IEEE 802.2 Profile for PC Networks (1 of 2) ] ]
]] ] ]
]] Adapter number . . . . . . . . . . . : n ] ]
]] ] ]
]] Use Universally administered address : Yes ] ]
]] Adapter address . . . . . . . . . . : ] ]
]] Functional address . . . . . . . . . : ] ]
]] Group address . . . . . . . . . . . : ] ]
]] ] ]
]] Maximum number SAPs . . . . . . . . : 2 ] ]
]] Maximum link stations . . . . . . . : 8 ] ]
]] Maximum number group SAPs . . . . . : 0 ] ]
]] Maximum members per group SAP . . . : 0 ] ]
]] Maximum number of users . . . . . . : 2 ] ]
]] ] ]
]] Transmit buffer size . . . . . . . . : 1048 bytes ] ]
]] Number of transmit buffers . . . . . : 2 ] ]
]] Receive buffer size . . . . . . . . : 280 bytes ] ]
]] Minimum receive buffers . . . . . . : 10 ] ]
]] ] ]
]] Adapter workarea size . . . . . . . : 16K ] ]
]] ] ]
]]----------------------------------------------------------------------] ]
]] Esc=Cancel F1=Help F8=Forward ] ]
]*----------------------------------------------------------------------* ]
] ]
] ]
]*----------------------------------------------------------------------* ]
]] LAN IEEE 802.2 Profile for PC Networks (2 of 2) ] ]
]] ] ]
]] Adapter number . . . . . . . . . . . . . . : n ] ]
]] ] ]
]] Group 1 Response Timer . . . . . . . . . . : 50 x 40 ms. ] ]
]] Group 1 Acknowledgement Timer . . . . . . : 50 x 40 ms. ] ]
]] Group 1 Inactivity Timer . . . . . . . . . : 255 x 40 ms. ] ]
]] Group 2 Response Timer . . . . . . . . . . : 50 x 40 ms. ] ]
]] Group 2 Acknowledgement Timer . . . . . . : 50 x 40 ms. ] ]
]] Group 2 Inactivity Timer . . . . . . . . . : 255 x 40 ms. ] ]
]] ] ]
]] Common LAN adapters queue size . . . . . . : 400 elements ] ]
]] Number Global Descriptor Table Selectors . : 50 ] ]
]] Attended mode of initialization . . . . . : No ] ]
]] ] ]
]]----------------------------------------------------------------------] ]
]] Esc=Cancel F1=Help F7=Backward ] ]
]*----------------------------------------------------------------------* ]
] ]
] ]
+----------------------------------------------------------------------------+
Figure 79. OS/2 EE 1.1 - IEEE 802.2 Profile for PC Networks
Relevant parameters not described earlier are explained below:
ADAPTER NUMBER AND VERSION
An integer indicating whether the adapter is primary (0) or
secondary (1) and the adapter type.
ADAPTER SHARED MEMORY
The shared RAM location in the workstation's memory.
GROUP ADDRESS
The group address is not used by the LAN data link control. It
needs only to be changed for other applications requiring a group
SAP address.
MAXIMUM NUMBER SAPS
LAN data link control requires one dedicated service access point
per adapter. Additional application using SAPs must be considered
in addition to this SAP.
MAXIMUM LINK STATIONS
The total number of link stations required by all LAN applications
concurrently using the LAN adapter. Some applications may share SAP
and link stations.
MAXIMUM NUMBER GROUP SAPS
Group SAPs are not used by LAN data link control. This number needs
to be changed only for other applications requiring a group SAP
address.
MAXIMUM MEMBERS PER GROUP SAP
Group SAPs are not used by LAN data link control. This number needs
to be changed only for other applications requiring a Group SAP
address.
MAXIMUM NUMBER OF USERS
LAN data link control is considered as one user of each active
adapter. Other users/applications need to be added.
TRANSMIT BUFFER SIZE
This value must indicate the largest transmit buffer size used by
any application using the LAN adapter.
NUMBER OF TRANSMIT BUFFERS
Due to limited shared RAM memory and the size of an SNA RU or
non-SNA messages, one or two is considered optimal. There should be
several receive buffers for each transmit buffer. There is some
performance advantage in overlapped processing and I/O on the
adapter for which two transmit buffers are required.
RECEIVE BUFFER SIZE
Ideally this value should be large enough to contain the average
size data frame received on the network. The default value may have
to be increased, depending on LAN application and data traffic
characteristics.
MINIMUM RECEIVE BUFFERS
A minimum of two receive buffers is required to open the adapter.
The remaining shared RAM memory, after all other storage
requirements have been met, will be used for receive buffers.
SYSTEM KEY (HEX CHARACTER)
Specification of a key permits protection of adapter resources (for
example, protection from conflict with other adapters)
RESPONSE TIMER
This timer is maintained by the sending adapter whenever an I-format
LPDU frame is sent with the "Final" bit set (see "3.2.2.3 LLC
Protocol Data Unit"). If this timer expires before a response is
received, the sending stations solicits remote link station status.
Therefore this value should be set greater than the total maximum
frame delays within the sending station, the network and the
destination station.
ACKNOWLEDGEMENT TIMER
This timer is started by a link station whenever an I-format LPDU
frame is received into workstation memory, and stopped when an
acknowledgement is sent either with an outgoing frame or when the
number of I-format LPDUs received before sending acknowledgement
(window size) is reached. If the timer expires, the link station
should send an acknowledgement as soon as possible. The timer value
should be shorter than the response timer to ensure that the remote
link can receive the delayed acknowledgement before the response
timer expires.
INACTIVITY TIMER
This timer runs whenever the response timer is not running.
Expiration of this timer suggests that the link station may be lost.
It should be set five to ten times greater than the response timer.
ADAPTER WORKAREA SIZE
The PC Network work area is used to maintain SAPs and link stations.
The default value is 16 Kbytes.
5.6.4 OS/2 EE 1.1 LAN Support versus IBM LAN Support Program V1.10
Figure 80 under heading "5.6.4 OS/2 EE 1.1 LAN Support versus IBM LAN Support
Program V1.10" compares IEEE 802.2 command support provided by Operating
System/2 Extended Edition V1.1 Communications Manager with the IBM LAN Support
Program V1.10 under IBM PC/DOS 3.3 or 4.0.
In Operating System/2 Extended Edition V1.1 the Command Control Block (CCB) is
changed by increasing the number of fields from 7 to 13 with a corresponding
increase in CCB size from 16 to 26 bytes. However, compatible offsets have
been maintained for existing fields.
The invocation of IEEE 802.2 services has changed as well. Instead of
Interrupt 5C, CALL LANCMDS (call to a dynamic link routine) is used. Data
structures should be aligned on even byte boundaries to avoid migration
problems.
*--------------------------------------------------------------------*
] Commands ] LAN Support ]Operating System/2]
] ] Program ] Extended Edition ]
]------------------------------+------------------+------------------]
] ] ] ]
]Common Commands ] ] ]
] BUFFER.FREE ] X ] X ]
] BUFFER.GET ] X ] X ]
] READ ] ] X ]
] READ.CANCEL ] ] X ]
] RECEIVE ] X ] X ]
] RECEIVE.CANCEL ] X ] X ]
] RECEIVE.MODIFY ] X ] X ]
] TRANSMIT ] X ] X ]
] ] ] ]
]DIRECT Interface Commands ] ] ]
] DIR.CLOSE.ADAPTER ] X ] X ]
] DIR.CLOSE.DIRECT ] ] X ]
] DIR.DEFINE.MIF.ENVIRONMENT ] X ] ]
] DIR.IMPL.ENABLE ] X ] ]
] DIR.INITIALIZE ] X ] X ]
] DIR.INTERRUPT ] X ] X ]
] DIR.MODIFY.OPEN.PARM ] X ] ]
] DIR.OPEN.ADAPTER ] X ] X ]
] DIR.OPEN.DIRECT ] ] X ]
] DIR.READ.LOG ] X ] X ]
] DIR.RSTORE.OPEN.PARM ] X ] ]
] DIR.SET.EXCEPTION.FLAG ] ] X ]
] DIR.SET.FUNCTIONAL.ADDRESS ] X ] X ]
] DIR.SET.GROUP.ADDRESS ] X ] X ]
] DIR.SET.USER.APPENDAGE ] X ] ]
] DIR.STATUS ] X ] X ]
] DIR.TIMER.CANCEL ] X ] X ]
] DIR.TIMER.CANCEL.GROUP ] X ] X ]
] DIR.TIMER.SET ] X ] X ]
] ] ] ]
]DLC Commands ] ] ]
] DLC.CLOSE.SAP ] X ] X ]
] DLC.CLOSE.STATION ] X ] X ]
] DLC.CONNECT.STATION ] X ] X ]
] DLC.FLOW.CONTROL ] X ] X ]
] DLC.MODIFY ] X ] X ]
] DLC.OPEN.SAP ] X ] X ]
] DLC.OPEN.STATION ] X ] X ]
] DLC.REALLOCATE.STATIONS ] X ] X ]
] DLC.RESET ] X ] X ]
] DLC.SET.THRESHOLD ] ] X ]
] DLC.STATISTICS ] X ] X ]
*--------------------------------------------------------------------*
Figure 80. OS/2 EE 1.1 and LAN Support Program 802.2 Commands
5.6.5 IBM LAN Support Program Version 1.2
The IBM LAN Support Program Version 1.2 includes all the function and
capabilities of the earlier releases of LAN Support Program. That is, it
provides interface code between applications written to the IEEE 802.2, IEEE
802.5 or NetBIOS interface, and token-ring and IBM PC Network attachment
cards.
In addition, version 1.2 extends support to include Ethernet DIX (Digital
Intel Xerox) Version 2 and IEEE 802.3 local area networks.
This means that IBM Personal Computers and Personal System/2 workstations
running the DOS operating system and attached to Ethernet LAN segments can
access stations such as SNA gateways or application processors attached to
token-rings. Connectivity to the token-ring networks would be via the IBM
8209 LAN Bridge, see "6.7.2 Token-Ring to Ethernet/IEEE 802.3."
The mechanism for providing this function is by support of the NETWORK DEVICE
INTERFACE SPECIFICATION , (NDIS).
5.6.5.1.1 Network Device Interface Specification: NDIS is an evolving
"industry standard" that was developed by 3Com and Microsoft as a means of
ensuring that applications, particularly server applications, could interface
to the wide range of Ethernet cards provided by different manufacturers. It
is a programming interface at the MAC layer, equivalent to the direct
interface in IEEE 802.5.
By using this "standard" programming interface, the characteristics of the
card are isolated from the application program. Hence the application can be
coded independently of the card installed.
The version of NDIS supported by the IBM LAN Support Program Version 1.2 is
known as NDIS 1.01.
Similar support is announced for the OS/2 Operating System, with OS/2 Extended
Edition Version 1.2.
The application interface support offered by LAN Support Program 1.2 is
summarized in Figure 81 under heading "5.6.5.1.1 Network Device Interface
Specification".
-------------------------------------------------------------------------------
*------------------------------------------------------*
] ]
] User Application ]
] ]
*------------------------------------------------------*
] ] ]
] ] V
] ] *--------------*
] ] ] ]
] ] ] NetBIOS ]
] ] ] ]
] ] *--------------*
] ] ]
] V V
] *----------------------------------------*
] ] ]
] ] IEEE 802.2 ]
] ] ]
] *----------------------------------------*
] ] ] ]
V V ] V
*--------------------* ] *--------------*
] ] ] ] ]
] IEEE 802.5 ] ] ] NDIS 1.01 ]
] ] ] ] ]
*--------------------* ] *--------------*
] ] ]
V V V
*--------------* *--------------* *--------------*
] Token-Ring ] ] PC Network ] ] Ethernet ]
] Adapter ] ] Adapter ] ] Adapter ]
*--------------* *--------------* *--------------*
IBM token-ring IBM PC Network Supported Ethernet
adapters for adapters for adapters
PC and PS/2 PC and PS/2
-------------------------------------------------------------------------------
Figure 81. LAN Support Program Version 1.2 - LAN Interface Support
The LAN Support Program V1.2 supports up to two adapters per workstation.
These adapters can be:
o Any IBM Token-Ring Network adapter
o Any of the existing IBM PC Network adapters
o A supported Ethernet adapter card for operating on the Ethernet DIX
Version 2 or IEEE 802.3 network. These are the:
- 3Com Etherlink II (Model 3C503)
- 3Com Etherlink/MC (Model 3C5203)
- Western Digital Ethercard PLUS (Model WDLAN-ERP-F001)
- Western Digital Ethercard PLUS/A (Model WDLAN-EP/A-F001).
The Ethernet cards require their NDIS device driver. The driver is usually
included on the diskette packaged with the card.
o MACWD.DOS is the driver for the Western Digital Ethercard PLUS and
Ethercard PLUS/A.
o ELNKII.DOS is the driver for the 3Com Etherlink II and ELNKMC.DOS for the
3Com Etherlink/MC.
5.6.6 OS/2 Extended Edition 1.2 LAN Support
OS/2 Extended Edition 1.2 introduced ETHERAND support, which enables IBM PCs
and PS/2s to attach to Ethernet/IEEE 802.3 networks and be able to communicate
with each other or host gateways using the IEEE 802.2 and NetBIOS protocols. A
similar implementation in PC/DOS has already been discussed in "5.6.5 IBM LAN
Support Program Version 1.2."
5.7 3174-Peer Communications RPQ 8Q0718
Many customers have a large installed base of 3270 equipment, screens and
printers, that are attached to 3174 Establishment Control Units via coax
cabling. This equipment has been traditionally used for accessing IBM
System/370 mainframes for interactive work; for example accessing databases or
program development. Today, many end-users are seeing benefits in the use of
programmable workstations for processing their data. At the same time, the
need for mainframe access is maintained.
Connecting a programmable workstation to a 3174 Establishment Control Unit has
been possible for some time. A suitable 3270 emulator program, such as the IBM
PC 3270 Emulator Entry Level, running in the workstation has provided the
mainframe access, but with this approach the end-users were unable to gain the
benefit of being attached to a LAN. They could not share data, applications,
and gateway access between each other. The solution was to attach the
workstations directly to the LAN. This meant that in many customer locations,
an investment in recabling was required, since the coax cabling used between
workstation and 3174 Controller was not suitable for token-ring use.
Many miles of this 3270 coaxial cabling are installed in customers premises.
To replace it involves considerable costs, and often major disruption to
end-user services, both factors which may make the exercise prohibitive.
Installing a structured cabling system, based on shielded twisted pair is a
sensible approach in a new or re-furbished building; less easy in an existing
coax-cabled location.
The 3174-Peer Communications Request-for-Price-Quotation (RPQ) allows
programmable workstations, connected by coaxial or unshielded twisted pair
cabling to IBM 3174 Establishment Control Units, to obtain the benefits of
being attached to a LAN, at the same time being able to access IBM hosts. If
the 3174 is itself attached to a token-ring, the workstations can communicate
with other stations on the token-ring. The RPQ is therefore able to act as a
migration aid for customers who for cost or cabling considerations are not
able to implement a full token-ring installation.
5.7.1 General description
The 3174-Peer Communication RPQ allows a LAN environment (3174-LAN) to be
established for workstations attached to 3270 wiring. Workstations use their
existing 3270 attachment cards, protecting customer investment. The RPQ, in
conjunction with the 3174 Workstation Peer Communication Support Program
(3174-WPCSP), provides full IEEE 802.2 support for the workstations, allowing
them to run applications written to the IEEE 802.2 or NetBIOS interfaces
peer-to-peer. These may include 3270 Emulators, APPC/PC or LAN
server/requester programs.
The 3174 is required to have a 3174 Type 3A Dual-Speed Adapter installed, and
the RPQ allows this adapter to act as a bridge between the 3174-LAN segment
and any establishment token-ring. Thus the stations on the 3174-LAN segment
can transparently access any other station as though they were attached
directly to the ring. This is shown in Figure 82 under heading "5.7.1 General
description".
-------------------------------------------------------------------------------
*---------* *---------*
] Host ] ] Appl. ]
] Gateway ] ] Process.]
*---------* *---------*
] ]
*--------------------------*
*-------------* ] ] *--------------*
] File Server ]----] Establishment Token-Ring ]---] Print Server ]
*-------------* ] ] *--------------*
*--------------------------*
]
]
*---------+----------*
] *----------------* ]
] ]3174-Peer Bridge] ]
] ] function ] ]
3174 x3R ] *----------------* ]
with ] ] ]
3174-Peer ] *------------* ]
Communication ] ] 3174 ] ]
RPQ ] ]" Peer LAN "] ]
] ] function ] ]
] *------------* ]
] ] ] ]
*-----+------+-------*
] ]
Coax Port Coax Port
#1 #32
] ]
] ]
*---------* ] ] *---------*
] PS/2 ] ] ] ] ] ] PC ]
] A ] ]---------------* *---------------] ] B ]
*---------* *---------*
A IBM 3174 Workstation Peer A
*-----Communication Support Program-----*
(5799-PHL)
IBM 3270 Connection card (Family 2)
IBM 3278/9 Emulation Adapter Card (Family 1)
-------------------------------------------------------------------------------
Figure 82. 3174-Peer Communication RPQ
With reference to the figure, "PS/2 A" is able to access "PC B" using LAN
applications, as well as the file server, print server, communications gateway
and application processor on the establishment token-ring. "PC B" has the same
capability.
Fixed function 3270 terminals (3278,3279,3192 etc.) attached to the 3174 are
handled in the usual way by the 3174 microcode. The RPQ only provides LAN
function to those workstations that have the 3174 Workstation Peer
Communication Support Program installed.
5.7.1.1.1 3270 Host Access: A variety of options are available, depending on
software, 3174 model, and whether or not the 3174 is attached to a real
token-ring. The 3174 is able to support the workstations as either 3174-LAN
devices or 3270 control unit terminal (CUT) / distributed function terminal
(DFT) devices. The 3174 recognizes which method is being used by analyzing the
protocol used on the coax when the station loads its software and connects to
the 3174.
If the 3174 Workstation Peer Communication Support Program is loaded as a
device driver in the workstation, the 3174 places the workstation on the
3174-LAN segment. If the workstation powers on and does not include the
3174-WPCSP as a device driver in its configuration, it may load a DFT or CUT
mode emulator and work as such a device. Therefore, a given workstation can
flip between 3174-LAN mode and CUT/DFT mode, depending on the software loaded.
It CANNOT use LAN and CUT/DFT concurrently.
If the workstation is considered to be on the 3174-LAN segment, then the
workstation software used for host access can be IBM Personal
Communications/3270 or IBM Workstation Program, configured as a SNA physical
unit (PU) together with logical units (LUs). Personal Communications/3270 may
also be used as a 3270 DFT emulator.
When workstations are attached to the 3174-LAN segment, they access the S/370
host as would a token-ring attached station. That is, they use an SNA host
gateway, such as the IBM 3745 or IBM 3174 Model 01L. If the 3174 to which
they are attached has the token-ring gateway feature installed, they can use
that 3174 as their gateway. Some possible configurations are shown in
Figure 83 under heading "5.7.1.1.1 3270 Host Access"
-------------------------------------------------------------------------------
*------* *------*
] Host ] ] Host ]
*------* *------*
] ]
*------*IBM 3174-01L *------*
] ]or IBM 3745 ] ]IBM 3745
*------*Gateway *------*
] ]
*----------* ]/]Telecommunications
]token-ring] ]link
*----------* ]
] ]
*--------* *--------*
] ]IBM 3174-03R ] ]IBM 3174-01R
]*------*]3174-Peer RPQ ]*------*]Token-ring gateway
]*------*] ]*------*]3174-Peer RPQ
*-+----+-* *-+----+-*
] ] ] ]
*-* *-*PC/3270 V1 *-* *-*PC/3270 V1
*-* *-*3174-WPCSP *-* *-*3174-WPCSP
1) The 3174 is attached to 2) The 3174 is link attached.
a real token-ring. It must have the token-ring
The workstations are gateway feature present.
seen as DSPUs of the The workstations are PUs
3745 or 3174 gateway. on a "multipoint" line.
*------*
] Host ]
*------*
] *-------*
*------* ] Host ]
] ]IBM 3745 *-------*
*------* ]]
] ]]
]/]Telecommunications *--------*
]link ] ]IBM 3174-01L
] ]*------*]Token-ring gateway
token- ] ]*------*]3174-Peer RPQ
ring *--------* *-+----+-*
*-----* ] ]IBM 3174-12R ] ]
] ]-]*------*]Token-ring gateway *-* *-*PC/3270 V1
*-----* ]*------*]3174-Peer RPQ *-* *-*3174-WPCSP
*-* *-+----+-*
*-* ] ]
*-* *-*PC/3270 V1 4) The 3174 is channel attached
*-* *-*3174-WPCSP to the S/370 host.
No local token-ring exists.
3) The 3174 is link attached and
acting as a gateway for stations
on the token-ring.
-------------------------------------------------------------------------------
Figure 83. Typical 3174-Peer Communications Host Connectivity
Scenarios 2 and 4 of Figure 83 under heading "5.7.1.1.1 3270 Host Access"
require that the 3174-Peer Communications machines have a Type 3A Dual-Speed
adapter installed even though there is no token-ring present. This adapter is
the only adapter supported by the 3174-Peer Communications RPQ and must always
be installed, even if there is no real token-ring present. It provides the
token-ring logical link control function within the 3174. If the 3174 is not
attaching to a real token-ring, the card does not have to be plugged into a
token-ring access unit; it works in a wrap mode using the cable provided with
the card. A patch to the RPQ microcode is required when working in this way.
5.7.2 Customization and Management
Stations attached to the 3174-LAN segment have token-ring MAC addresses as
would stations attached directly to a token-ring. The bridge function within
the 3174 has a bridge number, and the 3174-LAN segment a ring number. The
bridge provides full source routing support. However, the bridge is not
visible to the IBM LAN Manager or IBM LAN System Manager.
During 3174-Peer customization, values are assigned to reflect the:
o Individual MAC addresses of the 3174-Peer devices
o Bridge number of the 3174-Peer bridge function within the controller
o Segment number of the 3174-LAN segment
o Hop count of the 3174-Peer bridge
o Initial forwarding status of the 3174-Peer bridge
o Performance thresholds.
Management of the 3174-LAN segment is provided through the 3174 online test
menus. A CUT mode device is required to manage the 3174 Peer LAN segment.
The tests allow the operator to display status information and control
attachment of the 3174-Peer devices. The status and configuration of the
3174-Peer bridge, together with similar performance statistics to those
viewable on IBM token-ring bridges can also be obtained via the menu driven
interface.
5.7.3 Hardware/Software Requirements
The 3174-Peer Communication RPQ can be installed in 3174 models:
o 01L, 21L, 01R, and 51R with features 3026, 3030, or 3044
o 11L, 21R, 11R, and 61R with features 3026 or 3044
o 03R and 53R with feature 3030
o 13R and 63R
o 02R, 12R and 62R with features 3026 or 3044.
The RPQ is offered on a base of 3174 Configuration Support B Release 1.1 or
Release 3.
IBM 3174 Workstation Peer Communication Support Program (5799-PHL) is the
corequisite software RPQ that must be installed in the coax-attached
workstations. This program provides the same function as the IBM LAN Support
Program in the workstation, interfacing to either the IBM 3278/9 Emulation
Adapter Card or the IBM 3270 Connection Card. The program provides the IEEE
802.2 and NetBIOS application interfaces, together with the logical link
control required for token-ring operation. Applications believe that they are
working directly on the token-ring.
On the PC or PS/2, software supporting the 3174-Peer Workstation Communication
Support Program is:
o IBM Personal Communication/3270 V1.0
o IBM 3270 Workstation Program V1.12
o IBM Advanced Program-to-Program Communications (APPC/PC) V1.11
o IBM PC Local Area Network Program V1.32
o IBM DOS LAN Requester (included with OS/2 LAN Server V1.2)
5.7.4 Summary
The 3174-Peer Communications RPQ, in conjunction with the 3174 Workstation
Peer Communication Support Program allows DOS based Personal Computers and
Personal System/2 workstations attached by 3270 wiring to IBM 3174s to
communicate with each other, with token-ring attached devices and System/370
hosts as though they were attached directly to a token-ring.
The RPQ provides a solution for those customers who are unable, for whatever
reason, to implement a full token-ring LAN, yet wish to provide end users with
LAN capability.
The RPQ provides a cost effective way to migrate to a LAN environment using
existing installed controllers, 3270 cabling and workstation adapters.
The RPQ can be fitted to any model of the 3174 Establishment Control Unit
family, except Models 8xR, 9xR and 52R, provided that enough storage and the
requisite features are present.
The RPQ requires 3174 licensed microcode level B1.1 or B3 or above to operate.
5.8 IBM LAN Offerings Summary
The IBM Token-Ring Network is IBM's strategic LAN offering for the campus and
office environment. It offers extensive, continuously evolving device
attachment capability, including workstations, communications devices and
mid-range systems.
The 16 Mbps data rate capability, with announced support by large number of
attaching devices, stresses the importance of the IBM Token-Ring Network as a
viable backbone LAN in addition to high-speed LAN functions. The topology,
protocol, and performance characteristics that make the 16 Mbps Token-Ring
Network desirable as a backbone LAN today, also offer the potential for
ultimate coexistence and operation as cost-effective intermediate or
supporting rings in future FDDI backbone LANs.
Connectivity improvements to the IBM PC Network resulting from the ability to
bridge to Token-Ring Networks and to attach workstations to multichannel
networks using industry standard high-split channel distribution frequencies
make PC Network (Broadband) solutions more attractive in environments
requiring broadband multipurpose cable (such as concurrent video and data) and
broadband manufacturing LANs for which MAP 3.0 support is not yet available or
required.
The IBM PC Network Baseband is a low-cost LAN, though with recent IBM
announcements that include support for chaining of extender units as well as
support by the IBM PC Network Bridge Program, workstation access to both other
workstations on the same LAN and to token-ring connected devices is improved.
Attached workstations use the LAN Support Program to obtain support equivalent
to that provided to IBM Token-Ring Network and PC Network (Broadband)
workstations. This includes use of the IEEE 802.2 LLC interface and access to
both NetBIOS and APPC/PC (LU 6.2).
With the announcement of support for Ethernet and IEEE 802.3 attached PC and
PS/2 workstations, using LAN Support Program 1.2 in the PC/DOS environment and
OS/2 Extended Edition 1.2 and bridging possible between Ethernets,
token-rings, and PC Networks, a logical single LAN can be created from
segments with very different MAC protocols with huge connectivity potential.
With software support for all the base programming interfaces possible on all
adapter types, a consistent application base may be built.
6.0 LAN Segments Interconnection
6.1 Bridges, Routers and Gateways
Interconnection of LAN segments may be supported by one or more of the
following three approaches:
o A bridge
o A router
o A gateway.
What are the fundamental differences between these approaches? Perhaps the
easiest way to show the difference is by reference to the OSI model for data
communications as illustrated in Figure 84 under heading "6.1 Bridges,
Routers and Gateways" below.
-------------------------------------------------------------------------------
*-------------------* *-------------------*
] ] A ] ]
] Application ] ] ] Application ]
] ] ] ]
]-------------------] ] ]-------------------]
] ] ] ]
] Presentation ] ] ] Gateway ]
] ] ] ]
]-------------------] ] ]-------------------]
] ] ] ]
] Session ] Gateway] ] Session ]
] ] ] ]
]-------------------] ] ]-------------------]
] ] ] ]
] Transport ] ] ] Transport ]
] ] ] ]
]-------------------] A ]-------------------]
] ] ] ] ]
] Network ] ] A ] Network ]
] ] ] ] ] ]
]-------------------] ] ] ]-------------------]
] ] ] ] ] ]
] Data Link Control ] A ] Router] ] Data Link Control ]
] ] ] ] ] ] ]
]-------------------] ]Bridge ] ] ]-------------------]
] ] ] ] ] ] ]
] Physical ] ] ] ] ] Physical ]
] ] ] ] ] ] ]
*-------------------* V V V *-------------------*
-------------------------------------------------------------------------------
Figure 84. OSI Model - Bridge/Router/Gateway
A BRIDGE operates at layer 2 of the model. A bridge performs the function of a
Medium Access Control (MAC) relay, and therefore may be independent of any
higher layer protocol, (including the logical link protocol). It can
therefore be used to transport multiple higher layer protocols from one LAN to
another. In the simplest case, their use is to extend a LAN quickly and
easily, a step that may be needed for a variety of reasons. Some of these are
listed below.
One of the main advantages is that installing a bridge has little impact on
end-user communications. However, application and logical link timers may have
to be adjusted if a remote bridge is installed that uses a relatively slow
communications link.
A bridge only needs to do two things; to filter the frames arriving to decide
whether they need to be forwarded over the bridge, and then to forward them to
the other bridge port. This makes them relatively simple devices, hence fast
and efficient.
Bridges are available to interconnect LAN segments that have different Medium
Access Control (MAC) protocols. As will be seen later on in this chapter, the
IBM 8209 Token-Ring/Ethernet Bridge (see "6.7.2 Token-Ring to Ethernet/IEEE
802.3") and the IBM PC Network Bridge (see "6.6.3 IBM PC Network Bridge
Program") are able to do exactly that. But even when a MAC layer protocol
conversion is required, transparency to higher layer protocols is maintained.
If a connection-oriented protocol is being used, the logical link is
end-to-end between the communicating stations.
A ROUTER generally provides functions equivalent to the Network layer (layer
3). This means that it will only support particular networking protocols, for
instance TCP/IP, or NETBIOS. Therefore, when a router is installed to
interconnect segments, care must be taken to select a device that is able to
route the required protocols. Nowadays, routers are available that are able
to interpret a wide range of standard and proprietary protocols. Some are
capable of supporting multiple protocols within a single router station.
Whereas bridges are able to use the FLAT ADDRESS SPACE provided by stations'
MAC addresses, routers tend to rely on a hierarchical addressing structure so
that they can make their routing decisions. A good example of a hierarchical
addressing structure is provided by the telephone numbering system. Dialling
an international number from an office may mean entering many numbers, but
their function is to route, based on a hierarchy of telephone switching
exchanges.
For instance - 9-011-44-71-123-4567
9 PABX - Give me an outside line, route to a local telephone exchange
011 Local exchange - route me to the international exchange
44 International exchange - route me to the United Kingdom
71 United Kingdom - route me to London, etc, etc.
If we wanted to dial another city within the United Kingdom, then the London
code would be the number to change; up to that point the routing required
would be the same, and hence the number that would be dialled. The implication
of this is that when a router is added to the network, the hierarchy of
addressing must be understood, and the impact of a change in the addressing
structure on the applications assessed.
Thus routers must understand the addressing structure and take decisions on
whether, or how, to forward frames. This structure may have to be defined to
them, as in SNA subarea networking, or they might be able to learn it by means
of some protocol flowing between the routers, as in SNA APPN. The Internet
protocol, as used in TCP/IP, uses a structure that divides addresses into a
NETWORK ADDRESS and a HOST ADDRESS. The routers forward frames between the
connected networking segments on the basis of the network address until the
destination network is found.
Routers are very often used to interconnect LAN segments that are
geographically separated, and use wide area network services provided by the
PTTs. This means that a more limited bandwidth is available than that provided
within a local area network. To utilize this limited bandwidth effectively, it
is desirable to stop BROADCAST frames being routed over the WAN. Although
these frames may have little impact on LAN utilization, they can severely
impact WAN performance and throughput. Hence, routers will usually forward
traffic that is explicitly addressed to them, blocking uncontrolled broadcast
traffic.
The protocol used between routers can be proprietary, that is it can be
defined by the implementer. The choice may depend on what sort of wide area
network the router is to use between the various routing devices.
The GATEWAY is the final method of interconnecting LAN segments. A gateway is
a system function that supports more than one communication architecture or
network addressing scheme to permit connectivity and interoperability between
the devices in the attached environments.
Gateway products usually support address mapping from one network to another,
and may also provide transformation of the data between the environments to
support end-to-end application connectivity. Gateways usually link networks at
higher layers than bridges or routers, ranging from layer 3 to layer 7.
6.2 Bridged LANs
A prime reason for considering bridging LANs is to expand the networking
capability by:
o Increasing the geographic area covered by the total LAN
o Increasing the number of attached devices or wiring closets above that
supported on a single LAN segment
o Providing connectivity between stations attached to different LAN segments
so that segment differences (due to use of different MAC protocols, speeds
or frequencies) are transparent to higher layer protocols
o Increasing the bandwidth available to stations on a single segment by
splitting a LAN segment into one or more bridged LAN segments. In this way
fewer stations have to share the same common medium and access protocol
mechanisms.
Even when LAN stations use the same MAC protocol, there may be other physical
constraints preventing a station from joining a particular LAN segment. In the
case of the IBM Token-Ring, bridge interconnection is required when:
o The maximum number of stations attached to a single physical ring for a
given type of media has been reached (see "5.1.1.1 Cabling Components").
o Some stations are attached via unshielded telephone twisted pair wire
while all others are attached to data grade media twisted pair wire. It is
not an IBM recommendation that shielded and unshielded wiring be mixed
within a single physical ring.
For the IBM PC Network, bridging will be required when stations that require
to communicate are attached to broadband segments that use different channel
pairs to transmit and receive information (see "5.1.3 IBM PC Network
(Broadband) Components"), or it is required to bridge a PC Network (Broadband
or Baseband) segment to a token-ring or another PC Network segment.
LAN bridging provides autonomous network management entities ("9.0 LAN
Management and Recovery"), and may reduce installation and maintenance
required in fast growing LAN environments. Network availability may be
further improved by providing alternate connectivity paths across bridges.
6.2.1 Bridge Configurations
Many factors should be considered when configuring a LAN consisting of a
number of interconnected LAN segments, each with its own (perhaps different)
MAC protocol.
This section introduces different topology considerations for interconnected
LAN segments. Topologies associated with protocols for various single segment
LAN (segment) were described in "2.1.2 Network Topologies."
Bridged LAN topology can be influenced by any of the following:
o The number of workstations physically supported by a particular type of
LAN or cabling.
o Requirements to balance or alleviate heavy traffic associated with
particular applications over one or more interconnected LAN segments.
o Requirements for a geographical approach to interconnecting LAN segments
by associating segments with specific areas within a building or campus
layout.
o Desire to concentrate communications by connecting users with related
information needs within LAN segments, known as affinity groups.
o Requirements for performance, reliability and/or availability which may be
addressed through use of parallel bridges (16 under heading "6.2.1 Bridge
Configurations") or parallel routes (17 under heading "6.2.1 Bridge
Configurations"), providing both increased capacity and backup paths.
---Footnote---
(16) Parallel bridges are bridges interconnecting the same two LAN segments.
--------------
---Footnote---
(17) Parallel routes define multiple end-to-end paths for the same source and
destination LAN segments.
--------------
o Requirements to separate some LAN stations from others for security
purposes by using bridges to provide controllable connectivity paths
between secure segments and other stations.
o Requirements to access special function devices such as host gateways or
LAN servers which may be best satisfied through use of a backbone
topology.
It is evident that there is no "best solution" for every network. However,
general guidelines can be given based on the strengths and weaknesses of
general topologies, and possible performance expectations.
6.2.1.1.1 General Guidelines: When the number of workstations is small
(typically a network of fewer than 40 workstations) the distribution of
workstations within the network will be determined mainly by physical
considerations. The existence of departmental groups or affinity groups will
also be an important factor in selecting a particular physical layout, as will
the number and location of servers.
However, when the number of workstations grows, other factors (such as
increased traffic load versus capacity, availability, performance, management
and network expandability), increase in importance.
In the remainder of this design section, we will primarily consider bridges
interconnecting two token-ring LAN segments, although other segment topologies
and MAC protocols could equally apply except where noted.
o SERVER LAN CONFIGURATION
If two work groups or departments share similar files or printers, but are
attached to separate rings for other reasons, a server ring can be used to
reduce the costs of providing such resources as letter quality printers or
large disk storage. See Figure 85 under heading "6.2.1.1.1 General
Guidelines".
-------------------------------------------------------------------------------
*-----------* *-----------* *---*
] ] ] ]---] ]
] LAN ] *---* ] LAN ] *---* File and
] ]---] ]---] ] Printer
] A ] *---* ] B ] *---* Servers
] ] BRIDGE ] ]---] ]
*-----------* *-----------* *---*
]
*---*
BRIDGE] ]
*---*
]
*-----------*
] ]
] LAN ]
] ]
] C ]
] ]
*-----------*
-------------------------------------------------------------------------------
Figure 85. Server LAN Configuration. LAN A and LAN C are two separate
departments with little inter-communication. They share servers
attached to LAN B.
o FULLY-INTERCONNECTED AND LOOP CONFIGURATIONS
A fully-interconnected configuration (mesh) provides alternate paths from
each LAN segment to another. Should a bridge or path fail, traffic can be
routed through an alternate path, thereby increasing availability of the
server segment.
This solution tends to become impractical as the number of LAN segments
grows. For N LANs, one would require N(N-1)/2 bridges. Therefore, a loop
configuration as shown in Figure 86 under heading "6.2.1.1.1 General
Guidelines" can be an acceptable compromise between availability versus
cost and complexity (18 under heading "6.2.1.1.1 General Guidelines").
---Footnote---
(18) A loop configuration with three LAN segments is also fully-interconnected.
--------------
-------------------------------------------------------------------------------
*-----------* *-----------*
] ] ] ]
] LAN ] *---* ] LAN ]
] ]---] ]---] ]
] A ] *---* ] B ]
] ] Bridge ] ]
*-----------* 1 *-----------*
] ]
*---* Bridge *---* Bridge
] ] 4 ] ] 2
*---* *---*
] ]
*-----------* *-----------*
] ] ] ]
] LAN ] *---* ] LAN ]
] ]---] ]---] ]
] D ] *---* ] C ]
] ] Bridge ] ]
*-----------* 3 *-----------*
-------------------------------------------------------------------------------
Figure 86. Loop Configuration with Four LANs
o PARALLEL BRIDGE CONFIGURATION
Parallel bridges can address the problems of heavy traffic flows through
particular bridges and/or high availability requirements. While the
failure of one bridge would impact connectivity between LAN segments over
that bridge, sessions could be recovered via the parallel bridge. In
today's environment this would require that the application be
re-initialized and the new route be discovered during that initialization
process.
-------------------------------------------------------------------------------
*-----------* *-----------*
] ] ] ]
] ] *---* ] ]
] ]---] ]---] ]
] LAN ] *---* ] LAN ]
] ] Bridge ] ]
] ] 1 ] ]
] A ] *---* ] B ]
] ]---] ]---] ]
] ] *---* ] ]
] ] Bridge ] ]
*-----------* 2 *-----------*
-------------------------------------------------------------------------------
Figure 87. Parallel Configuration
o BACKBONE LAN CONFIGURATION
If growth is an important factor, a backbone LAN configuration can provide
the necessary flexibility. It also offers common server and gateway access
to a potentially very large number of LAN stations.
In a backbone configuration a number of LAN segments, sometimes referred
to as departmental LANs, are all connected to the same backbone LAN as
shown in Figure 88 under heading "6.2.1.1.1 General Guidelines". This
implies that between any two LAN stations attached to departmental LANs,
there is always a communications path that includes relatively few
bridges, whatever the size of the bridged LAN.
-------------------------------------------------------------------------------
*---------* *---------*
] Host ] ] Server ]
] Gateway ] ] station ]
*---------* *---------*
] ]
*------------------------------------------------*
] Backbone LAN ]
*------------------------------------------------*
*---* *---* *---* *---*
]B 1] ]B 2] ]B 3] ]B n] Bridges
*---* *---* *---* *---*
*-----* *-----* *-----* *-----*
]LAN 1] ]LAN 2] ]LAN 3]................ ]LAN n] Departmental LANs
*-----* *-----* *-----* *-----*
*---* *---* *---*
]B a] ]B b] ]B c] Bridges
*---* *---* *---*
*-----------------------*
] ]
] Duplex Backbone LAN ] or 3-LAN Loop configuration
*-----------------------*
-------------------------------------------------------------------------------
Figure 88. Backbone LAN Configuration
Because of the potentially high concentration of traffic on the common
backbone LAN, a LAN segment with stable performance characteristics and
possible higher bandwidth than the individual segments may be required.
As explained in "2.2.1 Basic CSMA/CD Concepts" and "2.2.2 Basic
Token-Passing Ring Concepts," a 4 or 16 Mbps Token-Ring Network may be a
more desirable choice for a backbone LAN segment than a CSMA/CD LAN
segment.
As an example, the backbone LAN in Figure 88 under heading "6.2.1.1.1
General Guidelines" could be a 16 Mbps IBM Token-Ring Network,
interconnecting a mixture of IBM PC Network (Broadband) LANs and 4 Mbps
IBM Token-Ring Networks.
Because of high availability, backup and/or capacity considerations, one
might consider implementing a duplex backbone, as a mirror image of the
first one.
6.2.1.1.2 Additional Design Considerations for Bridged LANs: When designing
a bridged LAN, one cannot ignore the bridge protocol implemented in the bridge
products themselves. Special attention may have to be paid to specific bridge
parameters.
The bridging technique discussed in "6.3.1 Transparent Bridging," does not
allow concurrently active parallel bridges or parallel routes.
For a more detailed discussion of LAN design considerations, please refer to
IBM MULTISEGMENT LAN DESIGN GUIDELINES (GG24-3398).
When using a source routing bridge protocol as presented in "6.3.2 Source
Routing," one may want to consider bridge parameters called SINGLE ROUTE
BROADCAST and HOP COUNT, to prevent flooding of the bridged network with
control frames during route resolution.
When selecting a particular configuration, the bridge performance itself
should be considered in addition to the above factors.
Bridge processing within a bridged local area network adds some degree of
overhead to the overall network throughput. The amount of bridge overhead
depends on several factors, some of which are listed below:
o FRAME SIZE: as the size of the frame increases, the bridge throughput
increases. Thus the largest possible frame size should be used, subject
to the constraints of applications, memory or buffer management.
o BRIDGE HOP DELAY: this is the elapsed time between the end of the receive
stage for a frame entering the bridge, and the end of the transmit process
for the same frame leaving the bridge through another port. This time is
seldom more than a few milliseconds, increasing as the frame size and LAN
traffic loads increase, but potentially more important when the
interconnected MAC protocols are different or run at different speeds.
END-USER PERCEPTION of application delay may or may not be affected by bridge
delays. As a general guideline, with slower applications, such as those
involving disk access or host access, bridge delays will not normally be
perceived. Other, faster applications such as program loads or
memory-to-memory copies, may be impacted when the frames must cross one or
more heavily used bridges.
The following factors minimize the impact of the above delays:
o Frames flowing through a bridge may be given a higher traffic priority
than normal LAN station frames, thus reducing the probability of bridge
congestion due to token-wait time.
o Route selection (based upon source routing) where more than one path is
possible, is based upon the fastest rather than the shortest path.
In normal workload conditions, bridge processing does not constitute a
bottleneck in a bridged LAN environment.
The IBM bridge products provide performance monitoring information to assist
in monitoring traffic flow and bridge performance. See "6.6 IBM LAN Bridges."
6.3 Bridge Standards
Two different approaches to LAN bridging have been developed within the IEEE
community. Both of these architectures will now be described, together with a
new approach to solving the problem of inter-operability between the
architectures.
IBM included in its Token-Ring Architecture, see IBM TOKEN-RING NETWORK
ARCHITECTURE REFERENCE a bridge architecture called SOURCE ROUTING, and
developed products implementing this approach. This approach has now been
included in the IEEE 802.5 Standard.
In parallel, the IEEE 802.1 subcommittee was working on a MAC bridge
architecture called TRANSPARENT BRIDGING, which currently has the status of a
Draft IEEE Standard. As a requirement of the IEEE 802 committees, source
routing must be capable of inter-operating with transparent bridging at the
MAC level.
6.3.1 Transparent Bridging
This section describes the transparent bridging protocol and its associated
spanning tree algorithm as drafted by the IEEE 802.1 subcommittee.
In this bridge architecture, there is a clear distinction between the actual,
physical bridged LAN topology and the simply connected active topology, which
reflects use of a spanning tree algorithm. Figure 89 under heading "6.3.1
Transparent Bridging" shows the physical configuration of a sample bridged LAN
as well as a possible active topology.
+------------------------------------------------------------------------------
]
] *-------------------------------------------------------*
] ] LAN Segment 1 ]
] *-------------------------------------------------------*
] *---* *---* *---*
] *-] P ]-* *-] P ]-* *--] P ]--*
] ] *---* ] ] *---* ] ] *---* ]
] ] BR5 ] ] BR1 ] *---* BR2 *---*
] ] *---* ] ] *---* ] *-] P ] ] P ]-*
] *-] P ]-* *-] P ]-* ] *-------------* ]
] *---* *---* ] ]
] ] ] ] ]
] *-------------------------------* ] *---------------* *---------------*
] ] LAN Segment 5 ] ] ] LAN Segment 3 ] ] LAN Segment 4 ]
] *-------------------------------* ] *---------------* *---------------*
] *---* *---* ]
] *-] P ]-* *-] P ]-* ]
] ] *---* ] ] *---* ] ] P = Bridge Port
] ] BR3 ] ] BR4 ] ]
] ] *---* ] ] *---* ] ] BRn = Bridge number n
] *-] P ]-* *-] P ]-* ]
] *---* *---* ]
] *-----------------------------------------*
] ] LAN Segment 2 ]
] *-----------------------------------------*
]
] BRIDGED LAN - PHYSICAL TOPOLOGY
]
]
] *-------------------------------------------------------*
] ] LAN Segment 1 ]
] *-------------------------------------------------------*
] *---* *---* *---*
] *-] P ]-* *-] P ]-* *--] P ]--*
] ] *---* ] ] *---* ] ] *---* ]
] ] BR5 ] ] BR1 ] *---* BR2 *---*
] ] *---* ] ] *---* ] *-] P ] ] P ]-*
] *-] P ]-* *-] P ]-* ] *-------------* ]
] *---* *---* ] ]
] ] ] ]
] *-------------------------------* ] *---------------* *---------------*
] ] LAN Segment 5 ] ] ] LAN Segment 3 ] ] LAN Segment 4 ]
] *-------------------------------* ] *---------------* *---------------*
] *---* *---* ]
] *-] P ]-* *-] P ]-* ]
] ] *---* ] ] *---* ] ] P = Bridge Port
] ] BR3 ] ] BR4 ] ]
] ] *---* ] ] *---* ] ] BRn = Bridge number n
] *-] P ]-* *-] P ]-* ]
] *---* *---* ]
] *-----------------------------------------*
] ] LAN Segment 2 ]
] *-----------------------------------------*
]
] BRIDGED LAN - ACTIVE TOPOLOGY
+------------------------------------------------------------------------------
Figure 89. Bridged LAN - Physical and Active Topologies.
In an active topology, frames are forwarded through those bridge ports which
are said to be in FORWARDING state. Other bridge ports do not forward frames
and are held in BLOCKING state. Bridges only connect LAN segments to which
they have attached ports in forwarding state. A port in blocking state may be
put in forwarding state, that is become part of the active topology, if bridge
components fail, are removed or are added.
In an active topology, one bridge is known as the ROOT BRIDGE. Each LAN
segment has a bridge port connected to it, called the DESIGNATED PORT, which
forwards frames from that segment to the root (bridge). The bridge to which
the designated port for a particular LAN segment belongs is called the
DESIGNATED BRIDGE. The root bridge is implicitly the designated bridge for
all the LAN segments to which it is connected. At any particular bridge, the
root port and the designated ports, if any, are the ports in forwarding state.
As an example, bridge 1 in the active topology sample presented in Figure 89
under heading "6.3.1 Transparent Bridging" has been selected as the root
bridge. Therefore it is the designated bridge for LAN segments 1 and 2.
Bridge 2 is the designated bridge for LAN segments 3 and 4, while bridge 4 is
the designated bridge for LAN segment 5. This structure can be represented as
a logical tree topology, called a SPANNING TREE, as shown in Figure 90 under
heading "6.3.1 Transparent Bridging".
+------------------------------------------------------------------------------
]
] *---------------------*
] ] Bridge 1 ]
] ] Id 42 R-P-C 00 ]
] ] *- - - -* *- - - -* ]
] *-]P-Id 01]-]P-Id 02]-*
] ]Cost 10] ]Cost 10]
] *-------* *-------*
] ] ]
] *---------------------------------* *--------------------------------
] ] LAN Segment 1 ] ] LAN Segment 2
] *---------------------------------* *--------------------------------
] ] ] ] ]
] *-------* *-------* *-------* *-------*
] ]P-id 01] ]P-Id 02] ]P-id 01] ]P-id 01]
] *-]Cost 05]-* *-----]Cost 10]-----* *-]Cost 05]-* *-]Cost 05]
] ] *- - - -* ] ] *- - - -* ] ] *- - - -* ] ] *- - - -*
] ] Bridge 5 ] ] Bridge 2 ] ] Bridge 5 ] ] Bridge 4
] ] Id 83 ] ]Id 97 R-P-C 10 ] ] Id 83 ] ] Id 57
] ] R-P-C 05 ] ]*- - - -* *- - - -*] ] R-P-C 05 ] ] R-P-C 05
] ] *- - - -* ] *]P-Id 01] ]P-Id 03]* ] *- - - -* ] ] *- - - -*
] *-]P-Id 02]-* ]Cost 05] ]Cost 05] *-]P-Id 02]-* *-]P-Id 02]
] ]Cost 05] *-------* *-------* ]Cost 05] ]Cost 05]
] *-------* ] ] *-------* *-------*
] ] ] ]
] *----------------------* *----------------------* *------------------
] ] LAN Segment 3 ] ] LAN Segment 4 ] ] LAN Segment 5
] *----------------------* *----------------------* *------------------
]
]
+------------------------------------------------------------------------------
Figure 90. Spanning Tree Sample
Abbreviations in this figure have the following meanings:
o Bridge:
- Id = Bridge identifier (unique throughout bridged LAN)
- R-P-C = Root path cost (least-cost path between root and bridge port)
o Port:
- P-Id = Port identifier (associated with each bridge port)
- Cost = Path cost (default value is related to the type of the attached
LAN segment)
The active topology is described in a FILTERING DATABASE, containing static
and dynamic entries. STATIC ENTRIES are entered under explicit management
control and contain MAC addresses for which filtering is specified as well as
a port map that specifies the state (forwarding or blocking) of each port.
The following parameters are examples of static entries which, amongst others,
determine the active topology:
o Unique bridge identifiers
o The path cost for each bridge port
o The port identifier associated with each bridge port.
DYNAMIC ENTRIES result from a learning process through communication with
other bridges or LAN stations to take into account physical LAN configuration
changes. Dynamic entries also change whenever the active topology changes
(such as the result of updated path cost values). A dynamic entry contains a
MAC address for which filtering is specified and a port number to which frames
with the specified destination address are to be forwarded.
The filtering database may be initialized with static entries from a permanent
database.
Bridges exchange BRIDGE PROTOCOL DATA UNITS (BPDUs) to propagate topology
information, thereby enabling computation of the active topology and resulting
spanning tree. A frame containing a BPDU has a "functional address" as its
MAC destination address. This forces all active bridges connected to the same
LAN segment as the sending bridge to copy the BPDU information. The
information carried in the BPDU supports selection of the designated port for
the LAN segment (the one with the lowest root path cost). At the same time,
it allows any bridge to select from its designated ports the one which is the
closest (that is, lowest cost) to the root as its ROOT PORT. The root bridge
is the bridge with the highest priority bridge identifier.
Thus selection of the active topology of a bridged LAN can be managed by
setting proper values for bridge identifier (priority component), for path
cost and for port identifier for each bridge port.
Because of propagation delays involved in communicating control information, a
sharp transition from one active topology to another one is not possible.
Therefore wait times are provided by the protocol as well as intermediate port
states. Consequently, apart from being in disabled, blocking or forwarding
state, a port can be in listening or learning state. A port in blocking state
must enter listening state, followed by learning state before entering into
forwarding state (that is before actually participating in frame relay).
A very important implication of transparent bridging with its particular
spanning tree algorithm, is that one cannot have concurrently active parallel
bridges (interconnecting the same two LAN segments) or even concurrently
active parallel paths (interconnecting the same source and destination LAN
segments).
6.3.2 Source Routing
Source routing refers to the route-resolution architecture and protocols as
implemented in the IBM bridge products, and as described in the IBM TOKEN-RING
NETWORK ARCHITECTURE REFERENCE.
A route through a multi-segment LAN can be described as a sequence of SEGMENT
NUMBERS and BRIDGE NUMBERS. A segment number - bridge number combination is
called a ROUTE DESIGNATOR.
Therefore, before looking at the route-resolution mechanisms used in source
routing, Figure 91 under heading "6.3.2 Source Routing" shows how ROUTING
INFORMATION is carried within a token-passing ring frame, and how route
designators are entered in the RI field.
+------------------------------------------------------------------------------
]
]
] <-----I----->
] *------------------------------------------------------------*
] ] SD ] AC ] FC ] DA ] SA ] RI ] ] FCS ] ED ] FS ]
] *------------------------------------------------------------*
] I/G ] ] ]
] *--------------------* ] ] ]
] ]X] Destination Addr ]-* ] ]
] *--------------------* ] ]
] *--------------------* ] ]
] ]X] Source Address ]---------* ]
] *--------------------* ]
] Routing ]
] Indicator <---------------------- 18 Bytes -------------------->
] <---2---> <-----2---->
] *--------------------------------------/---------------*
] ] ROUTING ] ROUTE ] ROUTE ] ] ROUTE ]
] ] CONTROL ]DESIGNATOR 1]DESIGNATOR 2] ]DESIGNATOR 8]
] *--------------------------------------/---------------*
] ] ]
] *---------------------------------* ]*- RN (16 - k) bits
] 0 7 0 7 *] *- unique Ring Nbr.
] *---------------* *---------------* *- IB , k bits
] ]B]B]B]L]L]L]L]L] ]D]F]F]F]r]r]r]r] *- Individual Bridge
] *---------------* *---------------* nbr., unique for
] ] ] ] ] ] parallel bridges
] ] ] ] ] *- reserved
] ] ] ] *- Largest Frame
] ] ] *- Direction bit
] ] *- Length of RI (5 bits)
] *- Broadcast Indicators
]
+------------------------------------------------------------------------------
Figure 91. Token Passing Ring Frame - Routing Information Field
Figure 92 under heading "6.3.2 Source Routing" lists the control bit settings
involved in source routing.
+------------------------------------------------------------------------------
]
]
] *------------------------------------------------------------------*
] ]Largest ] size ] comments ]
] ]Frame ](bytes)] ]
] ]--------+------ +-------------------------------------------------]
] ] B'000' ] 516 ] I-field, min. for LLC Type 1 operation ]
] ] B'001' ] 1500 ] largest frame size supported in ISO 8802-3 LANs ]
] ] B'010' ] 2052 ] suitable for 24 x 80 full screen data xfer ]
] ] B'011' ] 4472 ] largest frame size supported in FDDI draft and ]
] ] ] ] in ISO 8802-5 LANs with T(Token_Holding) = 9ms ]
] ] B'100' ] 8191 ] largest frame size supported in ISO 8802-4 LANs ]
] ] B'101' ] ] reserved ]
] ] B'110' ] ] reserved ]
] ] B'111' ] ] used in all-routes broadcast frames ]
] ]------------------------------------------------------------------]
] ]Broadcast]designator ] comments ]
] ]---------+-------------+------------------------------------------]
] ] B'0XX' ]Non-broadcast] used in local or specific route frames ]
] ] B'10X' ]All-routes ] frame transmitted along every route ]
] ] ] Broadcast ] in the network to the destination station]
] ] B'11X' ]Single-route ] only designated bridges will relay the ]
] ] ] Broadcast ] frame from one segment to the other ]
] ]------------------------------------------------------------------]
] ]Route ]length] comments ]
] ]Designator ](bits)] ]
] ]-------------+------+---------------------------------------------]
] ] Ring-Number ] 16-k ] k is the same for all bridges in one ]
] ] Individual- ] k ] multi-ring LAN, typically k = 4 (Bridge V1) ]
] ] Bridge-Nr. ] ] ]
] *------------------------------------------------------------------*
]
]
+------------------------------------------------------------------------------
Figure 92. Routing Information Field - Code Bits
SOURCE ROUTING is started at session connect time by the DLC_LAN_MGR (data
link control LAN manager) component of a station taking actions to RESOLVE THE
ROUTE
Route resolution is usually a two-stage process, whereby first the destination
station is searched on the LOCAL LAN SEGMENT. If the destination cannot be
located on the local segment, the destination will be searched on remote
segments connected via bridges.
o On-segment route determination: the source station sends a TEST or an
IEEE 802.2 XID command LLC protocol data unit (LPDU) to the address of the
destination without any routing information. The destination, if present,
responds with a TEST or XID response LPDU. If no response LPDU is
returned within a specific amount of time, the destination address is
assumed not to be on the local LAN segment, and the second stage of the
route discovery is initiated.
o Off-segment route determination: the source station immediately
retransmits the TEST or XID LPDU, indicating however the presence of
routing information by setting byte 0, bit 0 of the source address field
to 1 and appending an initial 2-byte RI field after the Source Address
field (see Figure 91 under heading "6.3.2 Source Routing"). At this
point the architecture provides a choice between two main dynamic route
discovery processes: ALL-ROUTES BROADCAST ROUTE DETERMINATION and
SINGLE-ROUTE BROADCAST ROUTE DETERMINATION.
6.3.2.1.1 All-Routes Broadcast Route Determination: In a TEST or XID command
LPDU, broadcast to all rings by the source station, the first two bits of the
RI field are set to B'10'. This triggers all bridges to copy the LPDU frame,
based on Hop Count limitations set at the bridge, and while forwarding it, to
complete the RI field with additional route designator information.
The destination station receives as many command LPDUs as there are available
routes, while any received frame contains in its routing information field the
precise route which has been followed. Any received command LPDU will be
returned as a response LPDU, setting the first RI-bit to B'0' (non-broadcast)
and another routing control bit, the direction bit, to B'1'. This forces any
response frame to flow back to the source station via the route as built in
the LPDU of the command received.
The source station selects the PREFERRED ROUTE from all returned response
LPDUs, and subsequent transmissions to the destination station follow the
preferred route. The destination learns the preferred route from the first
non-broadcast frame received from the partner source station. The preferred
route information between two stations remains valid for the duration of the
DLC session between both.
6.3.2.1.2 Single-Route Broadcast Route Determination: A value of B'11' in
the first two bits of the RI field, marks the TEST or XID command LPDU as a
Single-Route Broadcast frame. Depending on the setting of the single-route
broadcast bridge parameter (see "6.6 IBM LAN Bridges"), the bridge will decide
whether or not to copy and forward the frame (leaving the RI-field unchanged).
In this way, each LAN segment may communicate only one copy of the TEST or XID
command LPDU and the destination station will receive only one command LPDU.
The TEST or XID command LPDU is then returned as a response LPDU with the
first two RI-bits indicating all-routes broadcast. Again multiple response
LPDUs may be received by a source station, but in this case routing
information has been collected in the response flow rather than the command
flow.
As in the first procedure, the source station selects the PREFERRED ROUTE from
all returned response LPDUs.
Single route broadcast at the bridge level may be set on or off at bridge
initialization time by a (controlling) LAN Manager. If all bridges in a LAN
are configured for automatic single-route path maintenance the active topology
will be dynamically maintained by the bridges. Alternatively, the
single-route broadcast specification may be defined as a manual setting. If
so defined and the path becomes unavailable for any reason, the LAN Manager
must take appropriate action to (manually) define an alternate path by
reconfiguring the bridge's Single-Route broadcast setting.
The dynamic single-route broadcast path maintenance option uses a spanning
tree algorithm to maintain the status of each participating bridge. Section
"6.6.1 IBM Token-Ring Bridges" explains how this option is made available in
the IBM Token-Ring Network Bridge Program V2.0, IBM Token-Ring Network Bridge
Program V2.1, butler2. and IBM PC Network Bridge Program products.
6.3.3 Source Routing/Transparent Bridging Inter-operability
As stated earlier in this chapter ("6.3 Bridge Standards"), a requirement of
the IEEE 802.1 committee has been that source routing bridges/stations must be
capable of inter-operating with transparent bridges/stations, that is, bridges
and stations that are not capable of interpreting routing information carried
within frames. Proponents of the two schemes have argued the benefits of both
approaches to bridging, while the standards for their operation have been
developed by two sub-committees within the IEEE.
o Source routing. IEEE 802.5, moving towards an International Standard as an
addendum to ISO 8802.5
o Transparent Bridging. IEEE 802.1 Part D, now a Draft International
Standard, DIS 10038, and widely accepted.
Much discussion has centered around an annex to DIS 10038 that would solve the
problem of inter-operability, but progress has been slow.
One of the problems is that the implications of the two techniques are not
only seen in the design and operation of the bridge itself; but also in the
impact on the end-stations that need to communicate. For instance, stations
that wish to use source routing have to make decisions on when to send
Single-Route or All-Routes Broadcast frames, as well as interpreting the
routing information field within the frame. Not all stations do this,
particularly those that are attached to networks that traditionally use
transparent bridging.
There may be occasions when, of the two stations that need to communicate, one
will understand source routing, the other will not. The goal of
inter-operability is not achieved.
In the bridge machines themselves, transparent bridges maintain tables of MAC
addresses in order to make decisions as to when to forward frames. Source
routing bridges build the routing information field and then use it to make
their forwarding decision. A source routing bridge will not forward frames
that do not contain routing information; frames from non-source routing
stations will not carry such information.
At a March 1990 meeting of the IEEE 802 committees, IBM proposed the idea of a
Source Routing Transparent (SRT) bridge that would solve these problems. An
SRT bridge would be based on the concept of the transparent bridge, but would
include source routing capability as a tower on top of the base function.
An SRT bridge would form a single spanning tree with all other SRT and
existing "standard" transparent bridges, and if the spanning tree allowed,
would forward frames that did not contain a Routing Information field. This
would create a network identical in appearance and topology to that provided
by today's transparent bridges.
However, if a frame contained routing information, it would be forwarded in
the same way as that understood for the source routing bridges of today, even
if the bridge were in blocking state from the point-of-view of the spanning
tree.
SRT stations, stations that support both kinds of routing, would be able to
utilize a source routing path if it existed (all bridges between the stations
respective segments were SRT) or alternatively fall back on the spanning tree
path.
Figure 93 under heading "6.3.3 Source Routing/Transparent Bridging
Inter-operability" shows a network that is made up of transparent bridges
linking the LAN segments, and two workstations, X and Y, that need to
communicate. Neither X nor Y understand routing information, so therefore must
use the spanning tree path. This might create a bottleneck on bridges 1 and 2
as well as segment A, since all traffic between stations on segments B, C and
D will have to pass this way when bound for segments E, F, G and H.
There is of course, a much shorter path between segments D and F, via Bridge
8. However, the spanning tree has left this bridge in blocking state.
Therefore it will not be used to forward frames, until in this case, bridge 4
fails.
-------------------------------------------------------------------------------
*---*
]Seg]
] A ]
*---*
1*-* ]2
] *--------------*
*---* *---*
]Seg] ]Seg]
] B ] ] E ]
*---* *---*
] ] 5*---*]*--------------*7
3*--* *--*4 ] 6*----* ]
] ] *---* *---* *---*
*---* *---* ]Seg] ]Seg] ]Seg]
]Seg] ]Seg] -8- ] F ] ] G ] ] H ]
] C ] ] D ] *---* *---* *---*
*---* *---* ]
] *---* ] *---*
*--]W/S] *--]W/S]
] X ] ] Y ]
*---* *---*
All bridges (numeric) are transparent bridges.
LAN segments are denoted alphabetically.
Bridge 8 exists, but is in 'blocking' state.
W/S X and W/S Y do not "understand" routing information.
For W/S X, path to W/S Y is via D,4,B,1,A,2,E,5,F.
-------------------------------------------------------------------------------
Figure 93. Transparent Bridging
In Figure 94 under heading "6.3.3 Source Routing/Transparent Bridging
Inter-operability", Bridge 8 has been replaced by an SRT bridge. Although
still in blocking state according to the spanning tree, it is able to pass
frames that contain routing information. So, if the two workstations are able
to interpret and use routing information, that is are SRT stations, they can
now communicate on the direct path, D,8,F, saving the traffic overhead on the
previously intervening segments and bridges.
-------------------------------------------------------------------------------
*---*
]Seg]
] A ]
*---*
1*-* ]2
] *--------------*
*---* *---*
]Seg] ]Seg]
] B ] ] E ]
*---* *---*
] ] 5*---*]*--------------*7
3*--* *--*4 ] 6*----* ]
] ] *---* *---* *---*
*---* *---* ]Seg] ]Seg] ]Seg]
]Seg] ]Seg]--8--] F ] ] G ] ] H ]
] C ] ] D ] *---* *---* *---*
*---* *---* ]
] *---* ] *---*
*--]W/S] *--]W/S]
] X ] ] Y ]
*---* *---*
Bridges 1 - 7 are transparent bridges.
Bridge 8 is an SRT bridge.
Bridge 8 is in 'blocking' state as a transparent bridge, but will
forward frames that include routing information.
If W/S X and W/S Y both "understand" routing information, that is, are
SRT stations, then path is D,8,F. There is a source routing path.
Otherwise, route is D,4,B,1,A,2,E,5,F.
-------------------------------------------------------------------------------
Figure 94. An SRT Bridge in the Network
For today's source routing stations, minor changes in behavior would be
required to make them SRT stations. For instance, during route discovery, a
source routing station may transmit a Single-Route Broadcast frame. Releases
of the IBM Token-Ring Bridge Program Version 2, use a the standard
inter-bridge spanning tree protocol, to build a single-route broadcast path.
SRT bridges would be part of the "standard" spanning tree, as described by DIS
10038, so if a station transmitted a frame with no routing information (frame
now known as a ROUTE EXPLORER frame), that frame would arrive on each segment
of the LAN once. This achieves the same objective as the single-route
broadcast path, so therefore frames normally sent single-route broadcast would
simply be sent with no routing information and traverse the spanning tree.
Non-SRT stations would send the ROUTE EXPLORER response without routing
information, and future inter-station communication would be over the spanning
tree path.
SRT stations would transmit the response frame twice, once with no routing
information field, and once as a source routing All-Routes Broadcast frame.
At least one of these frames would get back to its destination, and the a
choice could be made on whatever route was available or most appropriate to be
used.
Group addressed frames currently sent single-route broadcast would be treated
the same way, that is sent without the routing information field present.
SOURCE ROUTING AND TRANSPARENT BRIDGING HARMONY IS OBTAINED.
It also means that all the benefits of source routing, such as parallel
bridges between segments, could be realized, without impacting the operation
of the spanning tree.
One of the other benefits of this approach is that one of the problems of
bridge and station inter-operability over other LAN types, such as FDDI, is
also solved. Any LAN implementation, present or future, needs only to add the
capability to support a routing information field in its MAC frame, and all
attaching stations, whether attaching directly or indirectly to that network,
supporting routing information fields or not, will be able to communicate
using one path or another.
It is no wonder this concept has been hailed as a breakthrough by all
concerned, and is rapidly moving forward within the standards arena.
By use of this approach, the best of both worlds is achieved. The two
techniques build upon each other. Customers will be able to build their
networks to use the benefits of source routing bridges where they feel
appropriate.
The techniques of the SRT bridge/station have little impact on the existing
"standards" work, in fact they make it considerably easier. However, it must
be remembered that the SRT bridging concept is only at the preliminary stages
of the "standardization" process. The enthusiasm with which the idea has been
received may mean that in the future, an SRT bridge product, implemented in
accordance with an international standard, may find its way to the market.
6.4 Bridge LAN Management Interface
IBM Token-Ring bridges have server components built into them that are used
for management purposes. These server components are able to use another
component called the LAN Reporting Mechanism (LRM) to establishing a
reporting link with up to four IBM LAN Managers or IBM LAN Network Managers.
Each reporting link is an IEEE 802.2 logical link control Type 2
(connection-oriented) link, dedicated to transport network management
information in either direction.
LAN management functions and products are discussed in a separate Chapter "9.0
LAN Management and Recovery." Some LAN management commands change the way in
which a LAN segment operates. When a bridge has a reporting link with several
IBM LAN Managers to control the same LAN segment, those commands must be
reserved to one LAN Manager only to avoid conflicts. This LAN Manager is
called the CONTROLLING LAN MANAGER for a given bridge. All other (up to three)
LAN Managers to which this bridge reports are called OBSERVING LAN MANAGERS.
The setting of a reporting link in the IBM LAN Management products is done at
initialization time. Any given IBM LAN Management product can have one, and
only one, reporting link specification.
The link between a bridge and its controlling LAN Manager is reporting link 0,
also referred to as authorization level 0. Once a reporting link 0 is
established, bridges reject subsequent attempts by other IBM LAN Managers to
establish a reporting link 0, thereby avoiding duplicate controlling LAN
Managers.
Observing LAN Managers have reporting links 1, 2 or 3 with a given bridge.
IBM Token-Ring bridges may receive requests from the IBM LAN Manager or IBM
LAN Network Manager, and send back appropriate responses over a reporting link
after executing the request on an attached ring segment.
Some of the commands can only be executed when they are received on reporting
link 0, that is from the controlling LAN Manager. These commands are:
o Set single-route broadcast in a bridge.
o Remove station on a ring segment.
o Set soft error logging options for a ring segment.
o Perform a path test (this command has been removed from LAN Network
Manager).
The communication between IBM LAN Management products and IBM Token-Ring
Network Bridge Program V2 is shown in Figure 95 under heading "6.4 Bridge LAN
Management Interface".
+------------------------------------------------------------------------------
]
]
] Observing Controlling Observing Observing
]LAN Managers *-------* *-------* *-------* *-------*
] ] ] ] ] ] ] ] ]
] ] . ] ] * ] ] . ] ] . ]
] *---.---* *---*---* *---.---* *---.---*
] Reporting Links 3 0 2 1
] (LLC Type 2) ........... * ..... .
] . **** . ............
]Bridge Program *-------------.-----*-----.-----.--------------*
] ] *------.-----*-----.-----.------* ]
] ] ] ] * ] ] * ] ] * ] ] * ] ] ]
] ] ] *---* *---* *---* *---* ] ]
] ] ] LAN Reporting Mechanism (LRM) ] ]
] ] *-------------------------------* ]
] ] ] ] ] ] ]
] ] *-----* *-----* *-----* *-------* ]
] ] ] REM ] ] RPS ] ] CRS ] ] LBS ] ]
] ] *-----* *-----* *-----* *-------* ]
] ] ] ] ] ] ]
] ] *-------------------------------* ]
] ] ] MAC Router ] ]
] ] *-------------------------------* ]
] ] ]
] ] *------------------------* ]
] ] ] Bridge Task ] ]
] *----------------------------------------------*
] ]Token-Ring Adapter] ]Token-Ring Adapter]
] *------------------* *------------------*
] Ring Segments ] ]
] *--------------------* *--------------------*
] ] -------------* ] ] -------* ]
] *----* ] ] ] ] ] ] *----*
] ] A ]----] Ring XXX ] ] ] Ring YYY ] ]---] B ]
] *----* ] ] ] ] ] ] *----*
] ] ] <-----* ] ] <-----* ] ]
] *--------------------* *--------------------*
] * - - - - - - - - LAN application LLC session - - - - - - - - *
]
]
+------------------------------------------------------------------------------
Figure 95. IBM Token-Ring Network Bridge Program V2 - IBM LAN (System) Manager
Communication
This figure summarizes both the application flow between end stations
(carrying data and higher-level-protocol header information), and network
management data flow. The latter consists of LLC Type 2 sessions between LAN
Manager and the LAN reporting mechanism which directs a LAN Manager
request/command to the appropriate server function in the Bridge. On the ring
segments side, LAN Manager requests/commands are translated into the
appropriate MAC frames for execution on one of the two attached ring segments.
Responses provided by the appropriate server function are directed to the LAN
reporting mechanism which packages them into an LLC frame addressed to the
originating LAN Manager.
Various server functions are also illustrated. Some of these, associated with
network management, are optional, while the LAN Bridge Server is always
present.
LAN BRIDGE SERVER
The bridge processing by the LBS function consists of:
o Reading and validating bridge parameters from a configuration file at
bridge initialization time and whenever a controlling LAN Manager modifies
bridge parameters.
o Performing the bridge self-test during bridge initialization or upon
request from an operator through the bridge user interface. This test
includes detection of duplicate parallel paths and invalid network
configurations (that is inconsistent ring numbering).
o Maintaining a set of performance counters for each adapter, including
counters for the number of frames discarded, not received or not forwarded
for any other reason, and for the number of frames and bytes forwarded
(both for broadcast and non-broadcast frames). On request, the
accumulated values may be reported to LAN Manager.
o Accumulating path trace information for frames carrying a system path
trace bit set on in the routing information control field. Path trace
report frames may be sent to the controlling LAN Manager.
The different bridge server functions which may be enabled optionally on each
of the bridge token-ring adapters are also shown in Figure 95 under heading
"6.4 Bridge LAN Management Interface".
RING ERROR MONITOR
The REM server function in the bridge compiles error statistics from received
Soft_ Error_ Report MAC frames and selectively generates reports for LAN
Manager, depending on the SOFT ERROR REPORTING MODE (normal/intensive) set by
LAN Manager.
RING PARAMETER SERVER
The Ring Parameter Server (RPS) is the target for all Request Initialization
MAC frames that are sent by ring stations during their attachment to the ring
segment. The RPS function makes the following parameters available to all
ring stations on the ring in response to the Request Initialization MAC frame:
o Ring Number
o Ring Station Soft Error Report Timer value (default of 2 seconds)
o Physical location (not currently implemented)
There can be more than one RPS function active on any give ring segment.
CONFIGURATION REPORT SERVER
The CRS function forwards configuration notifications to LAN Manager. Upon
receipt of a MAC level configuration notification, it transmits the received
information via the LAN reporting mechanism to LAN Manager. From LAN Manager,
CRS accepts such commands as Query Adapter, Remove Adapter and Set Station
Parameters for execution on a bridge-attached LAN segment.
6.5 Remote Bridge
A REMOTE BRIDGE, also called a split bridge, is a bridge which interconnects
LAN segments using a telecommunication link for the transmission of frames
between bridge ports, as conceptually shown in Figure 96 under heading "6.5
Remote Bridge".
+------------------------------------------------------------------------------
]
]*-----------* *-----------*
]] ] <------ Split Bridge ------> ] ]
]] ] *-----* ] ]
]] ] ]] ]]W1 *-----* ] ]
]] Ring N ]---] ]---------------/ ]] ]]P2] ]
]] ]P1]] ]] WAN/--------------] ]---] Ring M ]
]] ] *-----* W2]] ]] ] ]
]] ] *-----* ] ]
]*-----------* P1 = bridge port 1 ] ]
] P2 = bridge port 2 ] ]
] W1 = WAN communications port 1 ] ]
] W2 = WAN communications port 2 *-----------*
]
]
+------------------------------------------------------------------------------
Figure 96. Remote Bridge Concept
Physically, a remote bridge consists of two halves at each side of the
communications link. Instead of regular frame forwarding from one LAN adapter
to the other one as in a normal or local bridge, in a remote bridge a frame is
copied in one station from the LAN adapter to the communications adapter
(adding source routing information as required). This communications adapter
sends the frame across the communications link to the peer communications
adapter in the other bridge half, where the LLC frame is copied to the second
LAN adapter and transmitted on its LAN segment.
Functionally, a remote bridge provides logical link control level end-to-end
connectivity between LAN stations on either side of the bridge, supporting any
higher layer protocol carried over the LLC link. In contrast, a typical
remote gateway solution as offered by a remote System/370 host gateway device
supports only its own higher-level protocols, for example, LU1, LU2 and LU6.2,
between a LAN station and a System/370 main frame.
IBM provides remote bridge support in the IBM Token-Ring Network Bridge
Program V2.1 and V 2.2 products. This implementation is described in greater
detail in "6.6.1 IBM Token-Ring Bridges."
6.6 IBM LAN Bridges
IBM local area network bridges are implemented in dedicated machines attached
to two different LAN segments. For IBM Token-Ring Network Bridge Program V2
and IBM PC Network Bridge Program the bridge machine is a dedicated Personal
Computer or Personal System/2 workstation containing two LAN adapters,
(appropriate for the type of attached LAN segment), and a bridge program
running in the workstation's memory.
Since the bridge workstation is entirely dedicated to the bridging function,
these IBM Bridge Programs operate in a single-task IBM PC/DOS 3.3 or 4.0
environment.
The IBM 8209 LAN Bridge runs in stand-alone machines that do not require
PC/DOS support.
IBM bridges support source routing as described in "6.3.2 Source Routing."
Since the routing information field can only contain up to eight route
designators, end stations in a multi-segment environment communicate over a
maximum of seven source-routing bridges.
Communication across any IBM bridge is transparent to applications written to
the IEEE 802.2 logical link control.
The original bridge program, the IBM Token-Ring Network Bridge Program Version
1.0, is no longer marketed. This product supported interconnection of two 4
Mbps token-ring segments. It was succeeded by IBM TOKEN-RING NETWORK BRIDGE
PROGRAM V1.1, which also supports two 4 Mbps token-ring segments. In
addition, IBM Token-Ring Network Bridge Program V1.1 can communicate with up
to four LAN Managers implemented by IBM LAN Manager V1.0. Network management
aspects will be covered in greater detail in "9.1 IBM's Total Network
Management Strategy" .
The most recent IBM token-ring network bridge implementations are the IBM
Token-Ring Network Bridge Program V2.2 and IBM PC Network Bridge Program.
When all bridges in the network are running either IBM Token-Ring Network
Bridge Program V2 or IBM PC Network Bridge Program and are configured
appropriately, they will communicate with each other to automatically
configure the network single-route broadcast path. Refer to the single route
broadcast route determination algorithm in section "6.6.1.1.1.1 Single-Route
Broadcast Path Maintenance" for an introductory discussion of this
route-resolution technique.
IBM TOKEN-RING NETWORK BRIDGE PROGRAM V2.0
The IBM Token-Ring Network Bridge Program V2.0 interconnects
token-ring segments operating at either 4 Mbps or 16 Mbps. IBM
Token-Ring Network Bridge Program V2.0 communicates with up to four
LAN managers implemented by IBM LAN Manager V2.0.
IBM TOKEN-RING NETWORK BRIDGE PROGRAM V2.1
The IBM Token-Ring Network Bridge Program V2.1 was the first release
to provide REMOTE BRIDGING. It includes all the functions and
capabilities of IBM Token-Ring Network Bridge Program V2.0, and
provides bridging functions between remote token-ring segments (4
Mbps or 16 Mbps) over a leased teleprocessing (TP) line operating at
speeds from 9.6 Kbps to 1.344 MBps. In this case, a LAN workstation
at each end of the TP link will implement one half of the remote
bridge.
IBM Token-Ring Network Bridge Program V2.1 provides the capability
of filtering frames to be forwarded by the bridge. Via a
programming interface, a user program can determine which frames are
allowed to pass across the bridge. This feature is particularly
useful to limit bridge traffic when a relatively slow TP line is
used between the bridge halves (for example, 9.6 Kbps or 19.2 Kbps).
IBM TOKEN-RING NETWORK BRIDGE PROGRAM V2.2
The IBM Token-Ring Network Bridge Program V2.2 allows the use of a
dial up line between the two halves of a remote bridge, so that the
cost of a dedicated leased line for occasional traffic need not be
borne. This also means that the two LANs being interconnected by the
bridge can be varied according to the connectivity required.
IBM PC NETWORK BRIDGE PROGRAM.
A more general LAN segment interconnection solution is offered by
IBM PC Network Bridge Program. A bridge running the IBM PC Network
Bridge Program supports interconnection between any two of the
following LAN segments:
o 4 Mbps IBM Token-Ring Network
o PC Network (Baseband)
o PC Network (Broadband) operating on channel pair T13 - J
o PC Network (Broadband) operating on channel pair 2' - O
(Frequency 2)
o PC Network (Broadband) operating on channel pair 3' - P
(Frequency 3)
The PC Network segments, operating on different channel pairs, may
or may not share the same broadband medium.
In this way, PC Network attached devices may have access to host
gateway devices attached to token-ring segments of the network.
8209 LAN BRIDGE
The IBM 8209 allows either a token-ring to token-ring local bridge
or a token-ring to be bridged to a LAN segment that is running
Ethernet Version 2 or IEEE 802.3 MAC protocols. This dedicated
machine allows a rich set of inter-connection possibilities between
IBM and non-IBM equipment, given that many OEM products and
protocols are supported by Ethernet connectivity.
IBM Token-Ring Network Bridge Program V2, IBM 8209 LAN Bridge and IBM PC
Network Bridge Program are all capable of communicating with IBM LAN Manager
V2.0, and IBM LAN Network Manager, providing consistent network management for
the entire multi-segment LAN.
Before looking at the specific IBM bridge product implementations, Figure 97
under heading "6.6 IBM LAN Bridges" shows the major components and their
interfaces for a local bridge between two LAN segments.
+------------------------------------------------------------------------------
]
]
] *----------------------*
] LAN Manager <----->] Bridge Manager ]<-----> End-User
] *----------------------*
] ] Bridge Adapter ]
] ] Interface ]
] *-------------*<--->*----------------*<--->*-------------*
] ] LAN Adapter ] *--------* ] LAN Adapter ]
] ] microcode ]<------->] common ]<------->] microcode ]
] *-------------* ] memory ] *-------------*
] LAN Segment A *--------* LAN Segment B
]
]
+------------------------------------------------------------------------------
Figure 97. LAN Bridge Structure
Similarly, Figure 98 under heading "6.6 IBM LAN Bridges" illustrates the
major components and their inter-relationship for a remote bridge.
+------------------------------------------------------------------------------
]
]
]LAN Mgr <-->*----------------* *----------------*<--> LAN Mgr
] ] Bridge Manager ] ] Bridge Manager ]
]End-User <-->*----------------* *----------------*<--> End-User
] ]Bridge Adapter] ]Bridge Adapter]
] ] Interface ] ] Interface ]
]*-------*<--->*---------------* *---------------*<--->*-------*
]] LAN ] ]TP Comms.] ]TP Comms.] ] LAN ]
]]Adapter] ]Adapter ] ]Adapter ] ]Adapter]
]*-------* ]Interface] ]Interface] *-------*
]Segment A *---------* *---------* Segment B
] ] ]
] *---------* *---------*
] ]TP Comms.]-----/ ]TP Comms.]
] ]Adapter ] /-----]Adapter ]
] *---------* TP line *---------*
]
]
+------------------------------------------------------------------------------
Figure 98. Remote Bridge Structure
6.6.1 IBM Token-Ring Bridges
The IBM Token-Ring Bridge program versions 1.0 and 1.1 have been superseded by
IBM Token-Ring Network Bridge Program V2 which is available in three releases.
IBM Token-Ring Network Bridge Program V2.2 superseded IBM Token-Ring Network
Bridge Program V2.1, which was withdrawn from marketing in September 1990.
IBM TOKEN-RING NETWORK BRIDGE PROGRAM V1.1
o Interconnects 4Mbps ring segments
o Implements LAN Reporting Mechanism (LRM). See "6.4 Bridge LAN Management
Interface."
IBM TOKEN-RING NETWORK BRIDGE PROGRAM V2.0
o Supersedes IBM Token-Ring Network Bridge Program V1.1
o Interconnects 4Mbps and 16Mbps token-ring segments
o Implements DYNAMIC SINGLE-ROUTE BROADCAST PATH MAINTENANCE
6.6.1.1.1.1 Single-Route Broadcast Path Maintenance: The bridge can be
configured with the configuration utility or by the IBM LAN Manager V2.0 to
communicate with other active bridges to dynamically maintain the network
single-route broadcast path.
A bridge, configured to participate in the dynamic maintenance of the
network's single-route broadcast path, can be in any of three modes:
BLOCKING The bridge does not forward single-route broadcast frames and does
not participate in the bridge protocols.
LISTENING The bridge does not forward single-route broadcast frames but
participates in the bridge protocols.
FORWARDING
The bridge forwards single-route broadcast frames and participates
in the bridge protocols.
A bridge is in blocking mode during initialization. Once the bridge has
opened its adapters and has set the appropriate functional address, it enters
listening mode. After participating in the protocols long enough to determine
if it should forward single-route broadcast frames, the bridge will either
stay in listening mode or enter forwarding mode. Bridges participating in the
protocols monitor the single-route broadcast path with inter-bridge
communication, using Bridge Protocol Data Units (BPDUs). All inter-bridge
communication is sent as logical link control Type 1 (connectionless) data to
the bridge functional address (X'C00000000100'). The inter-bridge
communication is periodically initiated by a frame sent from the ROOT bridge.
This root bridge is automatically selected through processes specified by the
spanning tree algorithm.
The bridges will reconfigure the single-route broadcast path to accommodate
network changes caused by a new bridge being activated or an active bridge
being shutdown.
It is important to notice that single-route broadcast path maintenance does
not affect the way in which a bridge processes all-routes broadcast or
non-broadcast frames.
o Sends management information to IBM LAN Manager V2.
- Soft error reports and beaconing notification
- Bridge status and performance data
- Ring configuration reports.
IBM TOKEN-RING NETWORK BRIDGE PROGRAM V2.1
o Supersedes IBM Token-Ring Network Bridge Program V2.0
o Able to act as a remote bridge
IBM Token-Ring Network Bridge Program V2.1 can be configured as a REMOTE
BRIDGE, also called SPLIT BRIDGE.
A remote bridge consists of two bridge halves connected by a TP line.
Each bridge half requires a token-ring network adapter attached to a LAN
segment and a communications adapter. In this case, frames are
transferred between the two interconnected IBM token-ring network segments
via a leased teleprocessing (TP) line operating at speeds ranging from 9.6
Kbps to 1.344 Mbps.
Configuration information is passed across the communications link
directly during bridge initialization. Both bridge halves verify that the
common information matches (for instance bridge number and ring segment
number). In addition, when any of the operational bridge parameters is
changed by the IBM LAN Manager V2.0 or IBM LAN Network Manager, both
bridge halves will be updated.
Bridge performance reporting has also been enhanced to reflect split
bridge specific information. As mentioned earlier, a special FRAME NOT
FORWARDED, FILTERED counter is maintained to report the result of frame
filtering by a user exit. In addition, another counter is used to
accumulate the number of frames which are not forwarded because a cyclic
redundancy check (CRC) error is detected on the TP communications link.
Particularly when the TP link is operated at a relatively low transmission
speed, token-ring adapter buffering capacity may sometimes be insufficient
to store incoming frames from the ring segment before they can be
transmitted to the other bridge halve. Therefore, with a split bridge,
the FRAME NOT RECEIVED (ADAPTER CONGESTED) counter may reach its threshold
much faster than it would for the same segments and same traffic in a
standard bridge configuration.
The halves of the remote bridge can be attached to the telecommunication
network in one of the following ways:
- Via synchronous modems, providing the following interfaces at the
indicated speeds:
-- EIA RS-232C/CCITT V.24 at 9.6 Kbps to 19.2 Kbps.
-- CCITT V.35 at 9.6 Kbps to 1.344 Mbps.
-- X.21 bis/CCITT V.24 at 9.6 Kbps to 19.2 Kbps.
-- X.21 bis/CCITT V.35 at 9.6 Kbps to 1.344 Mbps.
-- X.21 (leased only) at 9.6 Kbps to 64 Kbps.
- Via a multiplexor, such as the Integrated Digital Network Exchange
(IDNX) Models 20, 40 and 70 through:
-- The USD or HSD communications adapter using CCITT V.35 at 9.6 Kbps
to 1.344 Mbps.
-- The QSD communications adapter using EIA RS-232C/CCITT V.24 at 9.6
Kbps to 19.2 Kbps.
-- The QSD communications adapter using CCITT V.35 at 9.6 Kbps to 56
Kbps.
An additional bridge parameter, COMMUNICATIONS ADAPTER ELECTRICAL
INTERFACE, identifies the electrical interface used by the communications
adapter to attach to the TP link. Valid options are 1 (RS-232), 2 (RS-422)
and 3 (V.35, default value).
The IBM Token-Ring Network Bridge Program V2.1 maintains an internal
protocol over the TP line between the two bridge halves.
Communications across a remote bridge, operating with a 1.344 Mbps TP
line, is transparent to applications written to the IEEE 802.2 logical
link control interface using source routing. However, in general,
different restrictions apply with respect to line speed, maximum time
delay, frame size, line error rate and number of active concurrent
sessions through the bridge. The common rational behind these different
restrictions is to avoid end-to-end session time-out due to excessive
delays experienced while crossing the remote bridge.
o Bridge frame filtering capability
FRAME-FORWARD FILTERING, provides a mechanism to limit the traffic across
a bridge. A bridge parameter, FILTERING ENABLED, enables invocation of a
FILTER USER APPENDAGE during the frame forwarding decision process. If
the filter appendage returns a no forward indicator the frame will be
discarded and a specific counter (frame not forwarded, filtered) is
incremented. A FILTER USER APPENDAGE program can be created through a
programming interface offered with the IBM Token-Ring Network Bridge
Program V2.1. Three filtering programs are supplied with the product. They
can be used to filter frames based on:
- Protocol, for example: do not pass NETBIOS frames
- Destination or source MAC addresses
- Number of logical link connections established over the bridge.
6.6.2 IBM Token-Ring Network Bridge Program Version 2.2
The IBM Token-Ring Network Bridge Program Version 2.2 enhances the function of
the IBM Token-Ring Network Bridge program V2.0 and V2.1 in the areas of remote
bridge support, while maintaining the same bridging and network management
functions as previous releases of Version 2 of the product.
For the remote bridge function, supported link speeds and the electrical
interfaces to the telecommunications link are identical to those supported by
IBM Token-Ring Bridge Program V2.1.
The new functions for remote bridge configurations are:
o Support for larger frames to cross the bridge
o Dial support, enabling the use of a low-speed analogue link using the
Public Switched Telephone Network (PSTN), between bridge halves.
o Provision for better management of the telecommunications link.
o Support for the new IBM 7820 Integrated Services Digital Network (ISDN)
Terminal Adapter, enabling bridge halves to be connected over an ISDN
network (also supported in IBM Token-Ring Network Bridge Program Version
2.1).
6.6.2.1.1 Larger Frame Sizes: In the IBM Token-Ring Network Bridge Program
V2.1, the first release of the IBM Token-Ring Bridge program to include a
"remote bridge" capability, the default value for the maximum frame size
supported was based on the speed of the link. There were three reasons for
this:
1. Transmitting long frames on slow links takes time, and therefore impacts
timer values used between communicating stations; for example, the logical
link control (LLC) acknowledgement timer (T1) may have to be lengthened to
take account of the increased transmission time.
2. Long frames could mean that the bridge becomes congested more easily.
Since the rate of frame transmission on the link could be much slower than
the rate of frame arrival from the ring, long queues of frames waiting to
be transmitted could develop in the bridge. Since applications are only
able to send a certain number of frames before requiring a response,
keeping the frames short means that individual stations cannot queue so
many bytes of data. This in turn means that the bridge should become less
easily congested, or that more concurrent connections can be efficiently
supported over the bridge.
3. Slower analog links are more prone to suffer errors than the higher-speed
digital links. Therefore the likelihood of a long frame on a slow link
suffering an error is increased. Since the remote bridge does not perform
any error recovery for link-related errors, recovery has to be done by the
end stations that are communicating over the bridge. This means that the
error rate must be kept low enough so that an application is not affected
by the re-transmission of its frames and the wasted time taken on the
link.
Maximum frame size also depended on the amount of storage allocated within the
bridge machine for the Communication Adapter Transmit Buffer Size. In the
original release of IBM Token-Ring Bridge Program 2.1, this was limited to
64Kb. This meant that if the allowable frame size was large, the number of
buffers available was restricted. A Program Temporary Fix (PTF) UR27835 was
made available to increase bridge throughput for small frame sizes. With the
PTF applied, high-speed links can now be driven at 85 percent for longer
periods of time.
NOTE: It should be pointed out that the default frame sizes were only
RECOMMENDATIONS. A token-ring administrator was able to override the defaults
when configuring the bridge program.
The largest frame size allowed when using the remote bridging capability of
IBM Token-Ring Bridge Program V2.1 was 2052 bytes. This value was recommended
when the link speed was greater than 56Kbps.
With IBM Token-Ring Bridge Program V2.2, the defaults are still related to
link speed, but the absolute maximum sizes have been increased to 8144 bytes
(16/16Mbps bridge only) or 4472 bytes (4/4Mbps or 4/16Mbps).
The maximum recommended (defaulted) frame sizes are detailed in Figure 99
under heading "6.6.2.1.1 Larger Frame Sizes".
+----------------------------------------------------------------------------+
] Figure 99. Recommended Maximum Frame Sizes ]
+--------------------------------------+-------------------------------------+
] TP Link Speed (Kbps) ] Frame size ]
+--------------------------------------+-------------------------------------+
] 9.6 to < 38.4 ] 516 ]
+--------------------------------------+-------------------------------------+
] 38.4 to < 56 ] 1500 ]
+--------------------------------------+-------------------------------------+
] 56 to < 1.344Mbs ] 2052 ]
+--------------------------------------+-------------------------------------+
] 1.344Mbps ] 4472 ]
+--------------------------------------+-------------------------------------+
Use of a frame size as large as 8144 is only possible when both sides of the
remote bridge have either the IBM Token-Ring 16/4 Adapter or the IBM
Token-Ring 16/4 Adapter/A installed with the full 64K of on-card RAM utilized,
both rings are operating at 16Mbps, and the TP link speed is 1.344Mbps. If a
high rate of errors is detected when using this value, the maximum frame size
should be reduced to 4472.
6.6.2.1.2 Cascading Remote Bridges: Cascading of remote bridges is not
restricted, however several factors must be considered when implementing this
configuration. The TP link speed is critical in determining whether
applications will work across cascaded remote bridges. The link protocol used
between the end stations, for example LLC Type 2, may have timer values which
must be met for the logical link to be maintained. In addition, the
application itself may have timing constraints. In a cascaded configuration,
and particularly at slower link speeds, these timing constraints may not be
met. It is the customer's responsibility to determine whether specific
applications will work in this configuration.
NOTE: As with local bridges, IBM Token-Ring Architecture restricts a frame
from crossing more than seven bridges in its path from end station to end
station.
6.6.2.1.3 Dial Support: Many customers need only occasional access to remote
LANs. For instance, a small stand-alone LAN, situated in a travel agency or a
banking branch, may provide all the local day-to-day services required.
Sometimes, perhaps for the purpose of software management or server
administration, there may be a need for skilled personnel to access the LAN
from a central site, or, for the branch office to transfer information to
another branch. A remote LAN bridge enables that function, but the cost of the
dedicated lines, used only occasionally, may make the approach expensive. By
using a switched telephone line between the bridge halves, connection can be
established when required, and also between different locations depending on
instantaneous need.
The IBM Token-Ring Bridge Version 2.2 can use a dial-up connection between the
two halves of a remote bridge. This support is provided by:
o New configuration parameters in the bridge halves
o Support of a V.25 bis modem, typically operating at 9.6Kbps, such as the
IBM 7855-10 V.32 modem
o Provision of an application program that will request the "local" half of
a remote bridge to make a call.
The scenario is illustrated in Figure 100 under heading "6.6.2.1.3 Dial
Support".
-------------------------------------------------------------------------------
*----* *---* *----* *---* *---*
]Srvr] ] ] ]Srvr] ] ] ] ]
*----* *---* *----* *---* *---*
*---------------* *------------------*
] Token-Ring "A"] ] Token-Ring "B" ]
*---------------* *------------------*
*---* *---*
] ]Bridge Half Bridge Half] ]
*---* *---*
*-----* *-----*
]Modem] ]Modem]
*-----* *-----*
&TELEPHONE. &TELEPHONE.
*-----*
]Modem]&TELEPHONE.
*-----*
*---*
] ]Bridge Half
*---*
*-------------* *-------------------------------*
] LAN ] ] ]
]Administrator]--] Head Office Token-Ring ]
*-------------* ] ]
*-------------------------------*
-------------------------------------------------------------------------------
Figure 100. Remote Bridge with Dial Support
The figure shows two small "branch-office" token-rings each with its own
server. The LAN administrator, using IBM Token-Ring Bridge Program V2.2 can
dial up either branch when required. Also Branch "A" is able to connect to
Branch "B" when they need to exchange data. The cost of the connection is
related to the length of the call, and the connectivity is versatile.
Dial support offers:
o Ability for a bridge to initiate outgoing calls, accept incoming, or both.
o Password protection, so that unauthorized users cannot initiate calls.
o Call setup using an IBM-provided application shipped with the product.
Users are also given information allowing them to write their own
application. This could allow the user a menu selection of possible
locations.
o Alternative ability to use a number pre-stored in the modem.
o Call progress signals from the bridge to the application program
indicating the status of the connection attempt.
o Call takedown when:
- Requested by the operator
- No frames have been forwarded over the bridge for a user-specifiable
time.
- Certain line error conditions occur.
The bridge is fully configurable for a wide range of timer values that may
be stipulated by country PTT regulations when operating V.25 bis modems on
switched circuits.
6.6.2.1.4 Telecommunications Link Management: Improved support for managing
a remote bridge is provided by the following:
o When displaying ring status information, both halves will display both
rings' status.
o When configuration information is displayed, both halves will display both
halves' configuration information.
o A new alert is provided to report the status of the link to the LAN
Network Manager Version 1.0, if the LAN Network Manager is linked to the
bridge.
Support is provided by the IBM LAN Network Manager Version 1.0 or later.
6.6.2.1.5 IBM 7820 ISDN Terminal Adapter Support: The IBM 7820 is a terminal
adapter that allows devices with the V.24, V.35 or X.21 interfaces to access
basic rate ISDN links. This means that the two halves of a remote bridge may
be connected over an ISDN network.
6.6.2.1.6 Hardware/Software Requirements: HARDWARE REQUIREMENTS:
o Computer
- IBM PC/AT
- IBM 7531, 7532, 7541, 7542, 7552, 7561 and 7562 Industrial Computers
- IBM Personal System/2 Model 30 or above.
o Minimum memory requirement is 512Kb.
o At least one 720Kb or 1.2Mb disk drive
o Token-ring adapters can be any of those supported by IBM Token-Ring Bridge
Program Version 2.1 or either of the IBM Token-Ring Network 16/4 Adapters.
o When running in the remote bridge configuration, two matching RTIC cards
are needed. Either:
- IBM Real-time Interface Co-processor Card with 512Kb storage (Family
1)
- IBM X.25 Interface Co-Processor/2 (Family 2)
together with the cable features to support the required electrical
interface - V.24, V.35, or X.21.
o If using the Dial support, a V.25 bis modem, such as the IBM 7855-10 V.32
modems
o If attaching to an ISDN Network, an IBM 7820 ISDN Terminal Adapter.
SOFTWARE REQUIREMENTS: For each half of a remote bridge configuration, the
following are required:
o IBM Token-Ring Network Bridge Program V2.2
o IBM PC DOS 3.3 or above
o IBM Real-time Interface Co-processor Support for DOS, Version 1 or higher.
6.6.3 IBM PC Network Bridge Program
The IBM PC Network Bridge Program provides MAC level interconnection between
the following LAN segments:
o Two IBM Token-Ring Network segments operating at 4 or 16 Mbps.
o Two IBM PC Network broadband segments.
o An IBM Token-Ring Network segment operating at 4 Mbps or 16 Mbps and an
IBM PC Network broadband segment.
Each individual IBM PC Network (Broadband) segment can operate at any of the
three supported frequency channel pairs. If PC Network (Broadband) segments
share the same broadband medium (at different frequencies), the IBM PC Network
Bridge Program product may be used to bridge between the two frequencies. In
this case, both adapters would be attached to the same piece of coaxial cable.
Although the IBM PC Network Bridge Program may interconnect two token-ring
network segments, its primary purpose is to integrate PC Network segments into
a larger (token-ring based) LAN environment, extending the connectivity
options for PC Network devices and allowing improved integration of LAN
management.
For any type of LAN interconnection, the IBM PC Network Bridge Program uses
the concepts of source routing. The bridge function is transparent to
applications written to the IEEE 802.2 logical link control interface using
source routing, whether the MAC protocol at either side of the bridge is
identical or dissimilar. The expected bridge throughput, however, depends on
the type of interconnected MAC segments.
The IBM PC Network Bridge Program includes all the network management support
provided by IBM Token-Ring Network Bridge Program V2 and also supports
forwarding of network management data from PC Network segments. Like IBM
Token-Ring Network Bridge Program V2, IBM PC Network Bridge Program interfaces
with up to four LAN Managers executing IBM LAN Manager V2.0 or IBM LAN Network
Manager.
The following IBM Token-Ring Network management information is sent to IBM LAN
Manager V2.0 or IBM LAN Network Manager:
o Soft error reports and beaconing notification
o Bridge status and performance data
o Ring configuration reports
o Path trace reports.
The following PC Network (Broadband) management information is sent to IBM LAN
Manager V2.0 or IBM LAN Network Manager:
o Continuous carrier and no carrier notifications
o Bridge status and performance data
o Ring configuration reports
o Path trace reports
o Topology reports (although the sequence between stations on a CSMA/CD LAN
segment is irrelevant).
The IBM PC Network Bridge Program also supports dynamic single-route broadcast
path configuration and maintenance.
Figure 101 under heading "6.6.3 IBM PC Network Bridge Program" shows the
general structure of IBM PC Network Bridge Program when connecting a PC
Network (Broadband) segment and an IBM Token-Ring Network segment. The IBM PC
Network Bridge Program includes adapter handlers for both PC Network and the
IBM Token-Ring Network, as well as logical link control and bridge protocol
software to support PC Network adapters.
+------------------------------------------------------------------------------
]
]
] *---------------------------------------------------------------------*
] ] IBM PC Network Bridge Program ]
] ] *-----------------------------------------* ]
] ] *--] LAN reporting mechanism (LRM) ]--* ]
] ] ] *-----------------------------------------* ] ]
] ] ] ] ] ] ] ] ] ] ]
] ] ] *---* *---* *---* *---* *---* *---* ] ]
] ] ] ]LEM] ]LPS] ]LBS] ]CRS] ]RPS] ]REM] ] ]
] ] ] *---* *---* *---* *---* *---* *---* ] ]
] ] ] ] ] ] ] ] ] ] ] ]
] ] ] *-------------------------------* ] ]
] ] ] ] CSMA/CD ] MAC Frame ] ] ]
] ] ] *-] Router ] Router ]-* ] ]
] ] ] ] *-------------------------------* ] ] ]
] ] ...].....]...................................].....]............... ]
] ] *-----------------------------------------------------* External ]
] ] ] PC Network ] Bridge Frame ] Token-Ring N/W ] Interface ]
] ] ] Adapter Handler ] Forward Logic ] Adapter Handler ] (Int X'5C')]
] ] ]-----------------+-----------------------------------* ]
] ] ] LLC ] Direct ] Bridge ] ] ] ] ]
] ] ]--------------------------] ] ] ] ]
] ] ] Protocol Interface Driver] ] ] ] ]
] ] *--------------------------* ] ] ] Hardware ]
] ] ] ] ] ] Interface ]
] *---------------------------------------------------------------------*
] ] ] ] ]
] *---------------------------------------------------------------------*
] ] Adapter ] ] ] ] ]
] ] Cards ] *-----------------------* ]
] ] ] ] Bridge ] Direct ] LLC ] ]
] ] *-----------------------* ]-----------------------] ]
] ] ] Medium Access Control ] ] Medium Access Control ] ]
] ] ]-----------------------] ]-----------------------] ]
] ] ] Physical Transport ] ] Physical Transport ] ]
] ] *-----------------------* *-----------------------* ]
] ] PC Network Adapter Token-Ring Network Adapter ]
] ] LAN segment A LAN segment B ]
] *---------------------------------------------------------------------*
]
+------------------------------------------------------------------------------
Figure 101. IBM PC Network Bridge Program - General Bridge Structure
The IBM PC Network Bridge Program end-user (19 under heading "6.6.3 IBM PC
Network Bridge Program") server functions, as opposed to internal system
server functions, are referenced explicitly in Figure 102 under heading "6.6.3
IBM PC Network Bridge Program".
]
]
]*-------------------------------------------------------------*
]] IBM PC Network Bridge SERVER ] Token-Ring ] PC Network ]
]] ] N/W Segment] Segment ]
]]-----------------------------------+------------+------------]
]] LAN reporting mechanism (LRM) ] Yes ] Yes ]
]]-----------------------------------+------------+------------]
]] Ring Error Monitor (REM) ] Yes ] No ]
]]-----------------------------------+------------+------------]
]] LAN Error Monitor (LEM) ] No ] Yes ]
]]-----------------------------------+------------+------------]
]] Ring Parameter Server (RPS) ] Yes ] No ]
]]-----------------------------------+------------+------------]
]] LAN Parameter Server (LPS) ] No ] Yes ]
]]-----------------------------------+------------+------------]
]] Configuration Report Server (CRS) ] Yes ] n/a ]
]]-----------------------------------+------------+------------]
]] LAN Bridge Server (LBS) ] Yes ] Yes ]
]*-------------------------------------------------------------*
]
]
+------------------------------------------------------------------------------
Figure 102. IBM PC Network Bridge Program End-User Servers
PC Network stations use LLC protocols for network control functions.
Therefore they can communicate directly with a LAN Manager, and PC Network
segments do not need a CRS server function provided by the bridge.
Bridge processing by the LAN Bridge Server function includes also SINGLE-ROUTE
BROADCAST PATH MAINTENANCE. If this spanning tree algorithm is desired, all
bridges in the multi-segment LAN must be configured to participate in the
dynamic maintenance protocols.
6.6.3.1.1.1 Hardware and Software Requirements: IBM PC Network Bridge
Program requires a dedicated IBM Personal System/2 Model 50, 60, 70 or 80. It
requires 384 Kbytes of memory and a 720 Kbyte or 1.2 Mbyte 3.5 inch diskette
drive.
Local area network adapters supported are shown in Figure 103 under heading
"6.6.3.1.1.1 Hardware and Software Requirements".
*------------------------------------------------------------*
] AFE ] P/N / FC ]
]-------------------------------------------+----------------]
] IBM Token-Ring Network Adapter /A ] 69X8138 / 4790 ]
] IBM Token-Ring Network 16/4 Adapter/A ] 16F1133 / 1133 ]
] IBM PC Network Adapter II/A ] 25F8278 / ]
] IBM PC Network Adapter II/A - Frequency 2 ] 25F8281 / ]
] IBM PC Network Adapter II/A - Frequency 3 ] 25F8284 / ]
]-------------------------------------------+----------------]
] (EMEA) ] P/N / FC ]
]-------------------------------------------+----------------]
] IBM Token-Ring Network Adapter /A ] 69X8138 / 4790 ]
] IBM Token-Ring Network 16/4 Adapter/A ] 16F1133 / 1133 ]
] IBM PC Network Adapter II/A ] 25F8280 / ]
] IBM PC Network Adapter II/A - Frequency 2 ] 25F8283 / ]
] IBM PC Network Adapter II/A - Frequency 3 ] 25F8286 / ]
*------------------------------------------------------------*
Figure 103. IBM PC Network Bridge Program - Bridge Supported Adapters
IBM PC Network Bridge Program's only software prerequisite is IBM PC/DOS 3.3
or 4.0 Devices attached to a broadband PC Network segment being bridged by the
IBM PC Network Bridge Program require IBM Local Area Network Support Program
Version 1.0 with PTF UR22583 or Version 1.1.
6.7 IBM 8209 LAN Bridge
The IBM 8209 LAN bridge is a stand-alone bridge product that may have one of
two configurations. Depending on the product feature code ordered, it may act
as a:
1. Token-Ring to Token-Ring bridge, when ordered with the IBM 8209 Token-Ring
attachment module
2. Token-Ring to Ethernet/IEEE802.3 bridge, when ordered with the IBM 8209
Ethernet or Enhanced Ethernet Attachment Feature.
The additional function of the Enhanced Ethernet Attachment feature is
described in "6.7.9 Enhanced Ethernet Attachment Feature."
As originally announced, the 8209 bridged an Ethernet/IEEE802.3 LAN to a
token-ring. The token-ring/token-ring capability is new. In either
configuration, one side of the bridge always connects to a token-ring.
The following pages describe the function and operation of both
configurations. As a token-ring to Ethernet bridge, the function is much more
complex, so the description is extended. A bridge between token-ring and
token-ring has already been described in relation to the IBM Token-Ring Bridge
Program, so the highlights and differences between the 8209 and the Bridge
Program will be all that is discussed here.
6.7.1 Token-Ring to Token-Ring
Configured in this way, the 8209 bridge:
o Can act only as a LOCAL bridge between two token-rings, with each ring
operating at either 4 or 16Mbps
o Is a stand-alone, rack mountable machine, requiring no keyboard or display
o Offers filtering and similar network management capabilities to the IBM
Token-Ring Bridge Program Version 2.2
o Provides PLUG AND PLAY customer installation
o Is supported for network management by the IBM LAN Manager V2.0 or above
o Is provided with a utility program that may be used to configure the
bridge and retrieve bridge profile if an IBM LAN Manager is not present on
the LAN.
Before the announcement of this product, token-ring bridges were provided by
software and PC or PS/2 hardware. This meant that the PCs or PS/2s required
keyboard and display components that were for the most part redundant. A
standalone bridge enables customers to have a black-box bridge function, and
allows them to re-use the the former bridge workstation as an end-user
machine.
The 8209 token-ring bridge implements all the network management functions
associated with the local IBM Token-Ring Network Bridge Program Version 2.2.
These include the functional addresses, (LAN Bridge Server, Ring Error Monitor
etc.) as well as the capability for single-route broadcast path maintenance.
A utility program, supplied with the token-ring attachment module allows the
user to set up filters to control the traffic forwarded between the rings, as
well as perform the bridge definition functions. The utility can be used to
set up the token-ring segment numbers, the bridge number and all the other
parameters usually defined within the IBM Token-Ring Bridge Program. Filtering
capabilities allow an address filter, as well as the setting up of customer
defined filters.
The 8209 token-ring to token-ring bridge should offer a performance
enhancement over the IBM Token-Ring Bridge Program implementation bridge
solution.
6.7.2 Token-Ring to Ethernet/IEEE 802.3
The IBM 8209 can provide an Ethernet/IEEE 802.3 to token-ring bridging
function when ordered with the Ethernet or Enhanced Ethernet Attachment
Feature. From discussions in previous chapters, in particular in "2.2 LAN
Medium Access Protocols" and "3.0 LAN Architectures and Standards," there are
many differences between the two types of LAN. They use different access
control mechanisms, CSMA/CD or token-passing, and even the name Ethernet is
commonly used to describe three different LAN types: Ethernet Version 1 (V1)
Ethernet Version 2 (V2) and IEEE 802.3.
Ethernet V1, now obsolete, differs from V2 at the physical layer, and Ethernet
V2, although compatible physically with IEEE 802.3, uses different frame
formats, such that stations conforming to one type can not correctly interpret
frames sent by the other. Token-ring frames are different from both.
Note that Ethernet Version 1 is not supported by the 8209.
Ethernet/IEEE 802.3 LANs have standardized on the transparent bridging
approach to interconnecting segments, using the spanning tree algorithm to
create a single network path while token-ring has opted for the source routing
architecture.
All this means that a protocol transparent, or MAC layer, bridge that
interconnects a token-ring segment with an Ethernet/IEEE 802.3 segment must do
more than receive a frame from one LAN segment, and transmit it on the other.
The IBM 8209 Token-Ring/Ethernet bridge is designed to perform the necessary
conversion of frames. The bridge must:
o Be seen by the token-ring segment as a source routing bridge
o Be seen by the Ethernet/IEEE 802.3 segment as a transparent bridge
o Support the IEEE spanning tree protocol
o Be able to interpret the frames of Ethernet Version 2 and map them into
token-ring frames (and vice versa)
o Be able to interpret the frames of the IEEE 802.3 protocol and map them
into token-ring frames (and vice versa).
The IBM 8209 is used to connect multiple LAN types into a single logical
network, as can be seen in Figure 104 under heading "6.7.2 Token-Ring to
Ethernet/IEEE 802.3".
-------------------------------------------------------------------------------
*---* *---* *---* *---* *------*
]W/S] ]W/S] ]W/S] ]W/S] ]Server]
*---* *---* *---* *---* *------*
-------------------------------------------------
] Ethernet V2 ]
] *----* *-------*
] ] *---* ]8209] ]Gateway]--/
E] *---*E/net to ]W/S] *----* *-------* /-----
t]-------] ]E/net Bridge *---* ] ]
h] *---* *---------------------*
e] ] Token-Ring ]
r] *---* *---------------------*
n]--]W/S] ]
e] *---* *-----*Token-Ring to
t] ]8209 ]Token-Ring Feature
] *-----*
V] *--* ]
2] ]8 ] *----------------------*
]--]2 ]----] Token-Ring ]
] ]0 ] *----------------------*
] ]9 ] ]
*--* *-----*Token-Ring to
]8209 ]Ethernet Feature
*-----*
]
------------------------------------------
] ] ] ] IEEE 802.3
*---* *---* *---* *---*
]W/S] ]W/S] ]W/S] ]W/S]
*---* *---* *---* *---*
*---*
]W/S] = Workstation
*---*
-------------------------------------------------------------------------------
Figure 104. Connecting Multiple LAN Types using the 8209
In this example, multiple 8209s have been used to connect token-ring attached
workstations to Ethernet V2 and IEEE 802.3 LANs, such that, given compatible
higher-layer protocols, workstations can communicate with each other, with the
server and the communication gateway. This extends the capabilities of both
sets of stations. It also means that bringing new LAN technology into an
established environment does not isolate the old and the new. Token-rings may
be installed alongside existing Ethernets, perhaps offering advantages in new
application areas, or token-rings can be used as backbones, connecting
multiple Ethernets together. The advantages of a token-ring as a backbone are
great. As discussed in "2.2 LAN Medium Access Protocols" , token-ring
architecture is capable of providing high data throughput under heavy load; a
characteristic that is valuable in a backbone LAN.
The IBM 8209 also lets an Ethernet backbone be used for carrying token-ring
traffic, though this configuration would mean that frames between token-rings
stations would be limited to lengths of less than 1500 bytes, the maximum
allowed on an Ethernet. This may be restricting for the applications, given
that token-rings can carry much larger frames.
As well as providing the customer with a homogeneous LAN "cloud", the IBM 8209
can also take advantage of the management architecture built into the IEEE
802.5 token-ring. This means that by implementing the functional richness of
the token-ring, and within the constraints of current Ethernet/IEEE 802.3
management, the LAN can be managed in its entirety.
6.7.3 General Description
The 8209 provides two LAN connections; one for token-ring and one for
Ethernet/IEEE 802.3. The token-ring port can run at either 4Mbps or 16Mbps.
The Ethernet port attaches by an Attachment Unit Interface (AUI), allowing the
8209 to be installed up to 50 meters from the transceiver tap on the coaxial
cable used as the Ethernet bus.
When ordered with the Ethernet attachment module, the IBM 8209 is shipped with
a utility program, suitable for running on a token-ring attached station. This
station may run either the OS/2 Extended Edition operating system, or PC/DOS
with the IBM LAN Support Program Version 1. The function of the utility
program will be described in "6.7.6 The Utility Program."
6.7.4 8209 Operation
Since Ethernet V2 and IEEE 802.3 are compatible at the physical layer, some
stations on a CSMA/CD segment may be using Ethernet frames while others use
IEEE 802.3. The different stations cannot communicate with each other, but
they can share the same physical medium. The 8209 can be set to recognize
only Ethernet V2 frames, only IEEE 802.3 frames, or automatically detect which
station is using which type of frame. This MODE SETTING depends on the
setting of the configuration switches. The modes are further described in
"6.7.5 Frame Conversion"; the switches in "6.7.7 Configuration Switches."
The 8209 maintains two databases internally; one for stations attached to the
Ethernet/IEEE 802.3 port, the other for token ring stations that have sent
frames to Ethernet stations. Entries in the Ethernet/IEEE 802.3 database may
be predefined (static), or dynamically created by the 8209. The static
entries are created by use of the 8209 utility program. The database for
token-ring stations contains only dynamically created entries.
The purpose of these databases is to contain addressing information for both
sets of stations, the frame formats used by them (Ethernet V2 or IEEE 802.3),
and source routing information for the token-ring. When the 8209 is powered
on and initialized, entries in the static part of the Ethernet/IEEE 802.3
database are initialized from pre-configured information kept in non-volatile
RAM. The 8209 then enters the LEARNING state and starts to learn the MAC
addresses of the stations on the Ethernet/IEEE 802.3 segment, as would any
other transparent bridge.
Suppose that an IBM 8209 has been installed between a token ring and an
Ethernet segment. Some stations on the Ethernet segment are using Ethernet V2
frames, others are using IEEE 802.3. The 8209 has been setup to automatically
determine which stations are using Ethernet or IEEE 802.3 frame structures.
The situation is depicted pictorially in Figure 105 under heading "6.7.4 8209
Operation".
-------------------------------------------------------------------------------
*----------* *----------* *----------* *----------* *----------*
]Mac=11111 ] ]Mac=22222 ] ]Mac=33333 ] ]Mac=44444 ] ]Mac=55555 ]
]Frme=802.3] ]Frme=E/net] ]Frme=802.3] ]Frme=E/net] ]Frme=E/net]
*----------* *----------* *----------* *----------* *----------*
] ] ] ] ]
--------------------------------------------------------------------
]
*------* Static Dbase. Dynamic Dbase.
] 8209 ] Mac Frme Mac Frme
*------* *-----------------* *-----------------*
] ]11111 802.3 ] ]22222 E/net ]
*-----------------* ]55555 E/net ] ]33333 802.3 ]
] ] ] ] ]44444 E/net ]
] Token-Ring ] ] ] ] ]
] ] *-----------------* *-----------------*
*-----------------* 8209 Ethernet/IEEE 802.3 databases
]
]
*----------* Mac Route
]Mac=66666 ] *-----------------*
] ] ] ]
*----------* ] ]
] ]
] ]
*-----------------*
8209 Token-Ring
database (dynamic)
-------------------------------------------------------------------------------
Figure 105. 8209 Databases - Learning
The Ethernet/IEEE 802.3 static database has been initialized from the
predefined entries. As these entries are kept in non-volatile storage, they
are not lost during an 8209 power outage. While the 8209 is learning,
stations on the Ethernet segment will be communicating with each other. The
8209 observes their frames, analyses them, and can update the dynamic database
with the stations' MAC addresses and the frame format. No entries are made in
the token-ring database.
After a predetermined time, and if the spanning tree algorithm allows, the
8209 enters the FORWARDING state. The 8209 keeps listening to the frames on
the Ethernet segment. Any frame that has a source address that is not in the
database will have an entry made. Thus, the 8209 continues to learn.
Now assume that the token-ring station with a MAC address 66666 wants to
communicate with the Ethernet station, address 22222. Station 66666 will send
out a frame, addressed to 22222, as a single route broadcast. This frame will
be copied by the IBM 8209, which will search its databases for an entry
corresponding to the destination MAC address 22222. Since the address and
protocol is known in the dynamic portion of the Ethernet/IEEE 802.3 database,
it will forward the frame in the correct format. The token-ring database will
be updated with the token-ring station's MAC address and any source routing
information. The IBM 8209 has a BRIDGE NUMBER , like any other token-ring
bridge, and the Ethernet is given an "emulated" SEGMENT OR "RING" number.
These values have been added to the diagram below, and the databases now
appear as in Figure 106 under heading "6.7.4 8209 Operation"
-------------------------------------------------------------------------------
*----------* *----------* *----------* *----------* *----------*
]Mac=11111 ] ]Mac=22222 ] ]Mac=33333 ] ]Mac=44444 ] ]Mac=55555 ]
]Pcol=802.3] ]Pcol=E/net] ]Pcol=802.3] ]Pcol=E/net] ]Pcol=E/net]
*----------* *----------* *----------* *----------* *----------*
] ] ] ] ]
--------------------------------------------------------------------
Segment=FF0]
*------* Static Dbase. Dynamic Dbase.
Bridge#1] 8209 ] Mac Pcol Mac Pcol
*------* *-----------------* *-----------------*
Ring=001 ] ]11111 802.3 ] ]22222 E/net ]
*-----------------* ]55555 E/net ] ]33333 802.3 ]
] ] ] ] ]44444 E/net ]
] Token-Ring ] ] ] ] ]
] ] *-----------------* *-----------------*
*-----------------* 8209 Ethernet/IEEE 802.3 databases
]
]
*----------* Mac Route
]Mac=66666 ] *-----------------*
] ] ]666666 001,1,FF0]
*----------* ] ]
] ]
] ]
*-----------------*
8209 Token-Ring
database (dynamic)
-------------------------------------------------------------------------------
Figure 106. 8209 Databases - Forwarding
When station 22222 returns a frame, the IBM 8209 is able to convert it from
Ethernet V2 format to token-ring format, adding the routing information,
extracted from the token-ring database.
For communication that originates on the Ethernet segment, the 8209 will
transmit the frame on its token-ring port after reformatting, provided that
the destination address does not exist in either of the Ethernet/IEEE 802.3
databases. This is standard transparent bridging technique - a transparent
bridge forwards a frame to its other port(s) if the destination address is
unknown to it. The frame will be transmitted on the token-ring as a broadcast
frame; whether single-route or all-routes depends on the type of frame. Any
response frame will be used to update the token-ring database with MAC address
and source routing information, since the destination station is now known to
be accessible through the token-ring.
The 8209 implements an AGING TIMER. This controls the length of time that
information remains in the dynamic databases. If no frames are seen for a
particular destination within this time, then the entry is deleted. This
prevents the databases becoming full, which would prevent new partner stations
communicating across the bridge. If an entry has "aged-out", it must be
rebuilt, using the same techniques that have been described.
6.7.5 Frame Conversion
So far, in reviewing the operation of the IBM 8209, we have stated that there
is a need for frame conversion between Ethernet/IEEE 802.3 LAN segments and
token-ring LAN segments, but we have not defined why this should be the case,
or what is required to be done. This section will show why this is necessary
and how it is done. Before continuing, the reader may want to refer to "3.2.2
Logical Link Control Sublayer." It should be remembered that:
o IEEE 802.5 Token-Rings use the IEEE 802.2 specification for logical link
control. 802.2 provides for both connectionless, TYPE 1 SERVICE , and
connection-oriented TYPE 2 SERVICE protocols.
o IEEE 802.3 CSMA/CD LANs also specify the use of IEEE 802.2 to provide the
logical link control, and mostly run connectionless.
o Ethernet V2 networks do not use logical link control. However, it can be
simulated by using additional software; for example, OS/2 EE 1.2
ETHERAND(20)
---Footnote---
(20) ETHERAND is a registered trademark of the IBM Corporation
--------------
or the LAN Support Program V1.2. See "5.6.5 IBM LAN Support Program
Version 1.2." This implementation is known as LLC-ON-ETHERNET.
This means that:
o Conversion from IEEE 802.3 to IEEE 802.5 is only done to map the frame
format - the 802.2 Protocol Data Unit (PDU) remains in both networks.
o Conversion from Ethernet V2 to 802.5 requires more processing.
Let us review the individual frame formats. These are shown in Figure 107
under heading "6.7.5 Frame Conversion"
-------------------------------------------------------------------------------
Ethernet V2
*-----------------------------------*
]Preamble]DA]SA]TYPE]Information]FCS]
*-----------------------------------*
IEEE 802.3
]-------802.2 LLC PDU---------]
*---------------------------------------------------------------*
]Preamble]SFD]DA]SA]Length]DSAP]SSAP]Control]Information]PAD]FCS]
*---------------------------------------------------------------*
IEEE 802.5 Token-Ring
]-------802.2 LLC PDU---------]
*---------------------------------------------------------*
]SD]AC]FC]DA]SA]RI]DSAP]SSAP]Control]Information]FCS]ED]FS]
*---------------------------------------------------------*
Key:
Common Ethernet V2 IEEE 802.3
DA Destination Address TYPE Upper layer SFD Starting Frame
SA Source Address Protocol used Delimiter
FCS Frame Check Sequence PAD Used to pad info.
to 46 bytes min.
Length - of LLC PDU
IEEE 802.5 IEEE 802.2 LLC PDU
SD Starting Delimiter LLC Logical Link Control
AC Access Control PDU Protocol Data Unit
FC Frame Control DSAP Destination Service Access Point
RI Routing Information SSAP Source Service Access Point
ED Ending Delimiter
FS Frame Status field
Control
This field contains the commands, responses, sequence numbers
and the Poll/Final bit, as used for Asynchronous Balanced Mode Extended.
Information
User data
-------------------------------------------------------------------------------
Figure 107. LAN Frame Formats
Apart from the 802.2 PDU present in an IEEE 802.3 frame, the most significant
difference between an IEEE 802.3 and an Ethernet V2 frame is the Length/Type
field.
6.7.5.1.1 Ethernet - Token-Ring
6.7.5.1.2 Sub-Network Access Protocol: The Internet community, those people
responsible for the development of the TCP/IP protocol (see "5.5 TCP/IP") have
developed a method for transferring data between IEEE networks (802.3, 802.4
and 802.5) and Ethernet V2 networks. This makes use of the SubNetwork Access
Protocol (SNAP). They have specified that for SNAP, the:
o The DSAP and the SSAP in the 802.2 header shall be set to x'AA'
o The control byte shall be set to x'03'
o The first three bytes of the information shall be x'00 00 00' (P_id)
o The next two bytes define the protocol being used.
This now allows a conversion between IEEE 802.5 Token-ring frames, and
Ethernet V2 frames. This process is summarized in Figure 108 under heading
"6.7.5.1.2 Sub-Network Access Protocol".
-------------------------------------------------------------------------------
Token Ring
]--------SNAP Header--------]
*-------------------------------------------------------------------*
]SD]AC]FC]DA]SA]RI]DSAP]SSAP]Control]P_id]Type]Information]FCS]ED]FS]
*-------------------------------------------------------------------*
] ]<--------Discard-------->] ]
]Copy ] ] Copy ]
V V V V
*--------------- - - - - - - - - - - - - ---------------------*
]Preamble]DA]SA] ]Type]Information]FCS]
*--------------- - - - - - - - - - - - - ---------------------*
Ethernet V2
Ethernet V2
*--------------- - - - - - - - - - - - - ---------------------*
]Preamble]DA]SA] ]Type]Information]FCS]
*--------------- - - - - - - - - - - - - ---------------------*
] ]<--------Insert--------->] ]
]Copy ] ] Copy ]
V V V V
*-------------------------------------------------------------------*
]SD]AC]FC]DA]SA]RI]DSAP]SSAP]Control]P_id]Type]Information]FCS]ED]FS]
*-------------------------------------------------------------------*
Token Ring
-------------------------------------------------------------------------------
Figure 108. Token-Ring - Ethernet V2 Frame Conversion
Thus, a protocol that uses SNAP headers can be successfully bridged between a
token-ring and an Ethernet V2 network. Such a protocol is TCP/IP.
6.7.5.1.3 LLC-on-Ethernet - Token-Ring: Logical Link Control protocols can
be carried in Ethernet frames. They are used for NETBIOS and SNA by the
PC-RT, OS/2 Extended Edition 1.2, and the LAN Support Program Version 1.2.
The information field of the Ethernet frame is formatted as in Figure 109
under heading "6.7.5.1.3 LLC-on-Ethernet - Token-Ring"
-------------------------------------------------------------------------------
Ethernet V2 Information field
*--------------------------------------------*
]Length]PAD]DSAP]SSAP]Control]Information]PAD]
*--------------------------------------------*
-------------------------------------------------------------------------------
Figure 109. LLC-on-Ethernet Information Field Format
In this frame format, the length value is computed to be from the DSAP value
to the last byte of the Information field, inclusive. The leading PAD is one
byte long, and the trailing PAD is only required if it is necessary to
increase the length to the minimum of 46.
Conversion from token-ring to Ethernet for the LLC-based protocols is
performed by the 8209 if the following conditions are met:
o LLC-on-Ethernet is enabled in the 8209 (the default is disabled).
The conversion is done when the 8209 recognizes a value of x'80D5' in the
Ethernet Type_Field, or a non SNAP-formatted token-ring frame is being
passed to an Ethernet station.
o The DSAP in the token-ring frame must also be a value defined in the 8209
SAP table. Default values in this table are x'00 04 F4 F0 08 and FC'.
These values correspond to the null SAP, (00), two SNA SAPs, (04 and 08),
the LAN management SAP, (F4), the NETBIOS SAP, (F0), and a Discovery SAP
(FC) used during Remote Program load. The process is shown in Figure 110
under heading "6.7.5.1.3 LLC-on-Ethernet - Token-Ring".
-------------------------------------------------------------------------------
Token-Ring
]--------802.2 LLC PDU--------]
*----------------------------------------------------------------------*
]SD]AC]FC]DA]SA] Routing Info. ]DSAP]SSAP]Control]Information]FCS]ED]FS]
*----------------------------------------------------------------------*
]Copy ] Discard ]<-----------Copy------------>]
] ] ] ]
V V Insert V V
*-----------------------------------------------------------*
]DA]SA]type]Length]PAD]DSAP]SSAP]Control]Information]PAD]FCS]
*-----------------------------------------------------------*
]80D5]<------Ethernet Information Field---------->]
Ethernet V2 ] ] See Figure 109 under heading "6.7.5.1.3 LLC-on-Ethern
Ethernet V2
]--------802.2 LLC PDU--------]
*--------------------------------------------------------------------*
]Preamble]DA]SA]type]Length]PAD]DSAP]SSAP]Control]Information]PAD]FCS]
*--------------------------------------------------------------------*
]Copy ] Discard ]<-----------Copy------------>]
] ] ] ]
V V Insert V V
*----------------------------------------------------------------------*
]SD]AC]FC]DA]SA] Routing Info. ]DSAP]SSAP]Control]Information]FCS]ED]FS]
*----------------------------------------------------------------------*
Token-Ring
The Ethernet Type field has a value x'80D5' to indicate LLC-on-Ethernet.
-------------------------------------------------------------------------------
Figure 110. LLC-on-Ethernet/Token-Ring Frame Conversion
6.7.5.1.4 IEEE 802.3 - Token-Ring: Token-ring to IEEE 802.3 conversion is
accomplished even more simply. See Figure 111 under heading "6.7.5.1.4 IEEE
802.3 - Token-Ring".
-------------------------------------------------------------------------------
Token-Ring
*-------------------------------------------------------------*
]SD]AC]FC]DA]SA] RI ]DSAP]SSAP]Control]Information]FCS]ED]FS]
*-------------------------------------------------------------*
] ] Cut ] ]
]Copy ] ]<-----------Copy------------>]
V VInsertV V
*---------------------------------------------------------------*
]Preamble]SFD]DA]SA]Length]DSAP]SSAP]Control]Information]PAD]FCS]
*---------------------------------------------------------------*
IEEE 802.3
IEEE 802.3
*---------------------------------------------------------------*
]Preamble]SFD]DA]SA]Length]DSAP]SSAP]Control]Information]PAD]FCS]
*---------------------------------------------------------------*
] ] Cut ]<-----------Copy------------>]
]Copy ] ] ]
V VInsertV V
*-------------------------------------------------------------*
]SD]AC]FC]DA]SA] RI ]DSAP]SSAP]Control]Information]FCS]ED]FS]
*-------------------------------------------------------------*
Token-Ring
-------------------------------------------------------------------------------
Figure 111. Token ring/IEEE 802.3/Token-Ring Frame Conversion
6.7.5.1.5 ARP and RARP: The 8209 also has the ability to convert frames that
conform to the Address Resolution Protocol (ARP) and the Reverse Address
Resolution Protocol (RARP). Conversion is not always necessary and depends
upon the interpretation of the protocol.
ARP frames are used by TCP/IP in request from a server a stations MAC address
when only its Internet address is known.
RARP is used by a station to find out its own Internet address from a server
machine. ARP and RARP packets have defined Ethernet type fields, (x'0806' and
x'8035' respectively); the contents of the information field are also defined.
6.7.6 The Utility Program
This program, shipped with every 8209 that includes an Ethernet attachment
module is designed to run under PC DOS (with LAN Support Program) or OS/2
Extended Edition. It allows the following unique 8209 parameters to be
defined, displayed and modified from a token-ring attached station:
o The Spanning tree parameters
o Operational mode - mode 1, mode 2, automatic
o Enable/disable early token release - token-ring port at 16Mbps.
o Filter Definitions - it is possible to setup the 8209 such that certain
frames will not pass across the bridge, thus reducing unnecessary traffic.
The most common use of this facility would be to filter on a protocol
basis; for example, only let TCP/IP frames pass across the bridge.
o Add/Delete Ethernet/IEEE 802.3 static database entries
o Examine Ethernet/IEEE 802.3 port statistics, for example the number of
frames in error, number of collisions.
o Define bridge and emulated ring numbers.
o Define mapping of Ethernet and token-ring addresses.
6.7.7 Configuration Switches
The IBM 8209 has five switches on the Ethernet attachment module. These are
used to set:
o Token-ring speed - may be set for 4Mbps or 16Mbps operation. If set for
16Mbps, then early token release is enabled.
o Enable/disable automatic mode selection. If enabled, the 8209 will
determine which frame format (Ethernet V2 or IEEE 802.3) a station on the
Ethernet/IEEE 802.3 port is using.
o Mode 1/mode 2 priority operation. The operation of these switches is
dependant on the setting of the Automatic Mode Switch.
MODE 1
Token-Ring to Ethernet V2
MODE 2
Token-Ring to IEEE 802.3
o Two switches: Initial Bridge number/Initial Ring number. These two
switches are used to set values in combination.
For further information the reader should consult the IBM 8209 CUSTOMER
INFORMATION MANUAL shipped with every machine.
6.7.8 Token-Ring Management Support
As has been mentioned, the IBM 8209 is viewed by token-ring stations as a
normal token-ring bridge. This also means that it is able to provide the
management functions of token-ring bridges. To that extent, the 8209
implements on its token-ring port the following functions:
o LAN bridge server
o Ring parameter server
o LAN reporting mechanism.
These server functions are described in "6.4 Bridge LAN Management Interface."
6.7.9 Enhanced Ethernet Attachment Feature
IBM has announced a new feature for the 8209 that increases the management
capability of the device for token-ring segments, particularly when
token-rings are being bridged to an Ethernet backbone. The new feature can be
installed in existing 8209s by replacing the old Ethernet attachment module
with the new feature.
o Ring Error Monitor (REM)
This enables the 8209 to receive MAC frames from token-ring attached
stations that are reporting errors. The information can be sent on to the
LAN Manager station on the network.
o Configuration Report Server (CRS)
The IBM 8209 now implements this functional address on its token-ring
port. Therefore, for configurations such as Figure 112 under heading
"6.7.9 Enhanced Ethernet Attachment Feature", the 8209 is able to report
token-ring stations attaching to the LAN.
-------------------------------------------------------------------------------
*-------------* *-------------*
] ] *------* *------* ] ]
] Token-Ring ]--] 8209 ]--* ] *--] 8209 ]-----] Token-Ring ]
] A ] ] A ] -------------] B ] A ] B ]
*-------------* *------* ] *------* ] *-------------*
] Ethernet A ]
*-----* ] CRS
] LAN ] <---------reporting link-------* REM
] Mgr ]
*-----*
8209 A and B have the Enhanced Ethernet Feature installed
-------------------------------------------------------------------------------
Figure 112. 8209 Ethernet Backbone - Enhanced Token-Ring Management
With the Enhanced Ethernet Feature, the Configuration Report Server
function is enabled on 8209 "B". Adapters accessing token-ring "B" are now
reported to the LAN Manager.
o Support of locally administered addresses on the 8209 ports.
o Enhanced Filtering
The filtering capability of the IBM 8209 is improved and provides the
ability to filter frames on Destination and Source Address, and a two byte
data field. You cannot filter frames on NETBIOS name or the number of
link stations being used.
o Additionally, IBM LAN Manager Version 2.0 or above is able to to maintain
a reporting link with 8209s across an Ethernet backbone by supporting use
of the LAN Management SAP, x'F4', through the existing LLC-on-Ethernet
function of the 8209.
o New functions are supported by an enhanced utility program.
6.7.10 Summary
The IBM 8209:
o When connected between two token-ring LANs:
- Is a standalone source-routing bridge
- Does not require a keyboard or display unit
- Has the function and filtering capabilities of the IBM Token-Ring
Bridge Program V2.2
- Is fully compatible with IBM LAN Network Manager
- Improves performance over the Token-Ring Bridge Program
o When configured for Ethernet/IEEE 802.3 to token-ring bridging:
- Is a standalone LAN bridge between an Ethernet/IEEE 802.3 and a
token-ring LAN.
- Is seen by stations on the Ethernet/IEEE 802.3 LAN as a transparent
bridge, and by stations on the token-ring as a source routing bridge
- Can pass user information transparently between the LANs, providing
frame conversion where necessary
- Supports communication between stations that use the Sub Network
Access Protocol, LLC-on-Ethernet, or the IEEE 802.2 Logical Link
Control.
These protocols are used by TCP/IP, OSI, SNA and NETBIOS. Access from
Ethernet/IEEE 802.3 LAN attached stations to IBM AS/400, IBM RS/6000
and IBM S/370 main frames (via gateways) is enabled.
- Allows peer-to-peer communication between workstations attached to
both LANs. IBM provides support for Ethernet/IEEE 802.3 attached IBM
Personal Computers and IBM Personal Systems/2 workstations in OS/2
Extended Edition 1.2 with Etherand support, LAN Support Program 1.2.,
as well as in the RS/6000 RISC range of processors and the IBM 6150
PC-RT, using SNA, TCP/IP and NETBIOS.
- Means that token-ring technology can be used to provide high speed,
high throughput, LAN backbones, providing an efficient, high
performance, single image local area network.
- Can be managed using the architected management protocols of IEEE
802.5, and hence by the IBM LAN Manager Version 2.0 or above. Full
token-ring management capability is available in conjunction with the
Enhanced Ethernet Feature.
- Is shipped with a utility program that allows setting of all the
bridge parameters.
6.8 IBM Bridge Products Coexistence and Migration
The IBM Token-Ring Network Bridge Program Version 1.0 and the IBM Token-Ring
Network Bridge Program V1.1 can coexist in a multi-segment network with the
IBM Token-Ring Network Bridge Program V2.0, the IBM Token-Ring Network Bridge
Program V2.1, the IBM Token-Ring Network Bridge Program V2.2, the IBM 8209 LAN
Bridge and the IBM PC Network Bridge Program.
However, to take advantage of the dynamic maintenance capability of the
single-route broadcast path, all bridges (with the exception of IBM 8209 LAN
Bridge) in the multi-segment network must be running either the IBM Token-Ring
Network Bridge Program V2 or the IBM PC Network Bridge Program, since only
those bridge products can be configured for automatic maintenance.
The IBM Token-Ring Network Bridge Program Version 1.0, now withdrawn from
marketing, has no functional capability to communicate with a LAN Manager.
The IBM Token-Ring Network Bridge Program V2.0, IBM Token-Ring Network Bridge
Program V2.1 IBM Token-Ring Network Bridge Program V2.2, IBM 8209 LAN Bridge
and IBM PC Network Bridge Program all communicate exclusively with IBM LAN
Manager V2.0 and IBM LAN Network Manager, providing extended network
management capabilities as explained in "9.0 LAN Management and Recovery."
One IBM LAN Manager V2.0 can maintain up to 64 concurrent reporting links with
any combination of these bridge products, while an IBM LAN Network Manager can
have a link to 255 bridges.
In today's large multi-segment LANs, full management capability, coupled with
the greatest connectivity is offered by the latest releases of the bridge and
LAN Management programs. This means that IBM Token-Ring Network Bridge Program
V2.2, IBM PC Network Bridge Program, IBM 8209 LAN Bridge and IBM LAN Network
Manager should become the products of choice for customers. Migrating these
products to new releases, as and when they are announced, should become an
accepted customer strategy.
6.9 Routers
6.9.1 The IBM LAN to LAN Wide Area Network Program
Since the early days of the IBM PC Network (Broadband), application programs,
such as the IBM PC Network Program and later the IBM PC LAN Program have
provided file-sharing and print-serving capabilities as well as requester
functions. The protocol that they used to communicate with each other over the
LAN was NETBIOS, which also provided a simple "session layer" programming
interface. NETBIOS is now accepted as an industry standard, with many hundreds
of PC/LAN applications written to use it. NETBIOS is still the protocol that
is used to connect the latest high-performing IBM LAN Server Program Version
1.2 to its requester machines. Although NETBIOS is no longer the IBM strategic
interface or protocol for use in today's distributed data and application
environment, the need to support it both on the local LAN and between
geographically separate LANs remains strong. Further information on NETBIOS
can be found in "5.6.1.2 NetBIOS Interface."
The IBM LAN to LAN Wide Area Network Program is a ROUTER product that is
designed to link geographically separate LANs together to carry NETBIOS
traffic between them. When the NETBIOS stations attach to LANs in close
geographical proximity, local bridges provide the most flexible approach to
network design. Other protocols besides NETBIOS are able to share the common
resource of the bridge.
When the LANs are too far apart to make use of local bridges, the question
arises of what should be done to provide interconnection? Many customers
already have a wide area network (WAN) installed, based on either SNA or X.25
switches. A WAN is usually designed to be a shared resource, and resulting
cost savings may mean that "meshing" becomes an economic proposition. This
approach has other significant benefits, such as providing multiple paths
through the network, and hence resilience to failure of individual components.
Could the LANs be linked over the existing WAN?
A remote bridge is another option. However, it requires the installation of a
dedicated telecommunications link between the LAN sites. This link cannot be
shared with the SNA WAN facilities, and to "mesh" the various LAN sites may
require many remote bridges with consequent costs and management implications
for the associated links. The IBM token-ring remote bridge implementation
does not allow for the link to cross a Packet Switched Data Network (PSDN)
accessed with the X.25 protocol, a common WAN implementation. The IBM LAN to
LAN Wide Area Network Program provides an answer to the question by allowing
NetBIOS communications between LANs to share an SNA or X.25 WAN.
6.9.1.1.1 General Description: For the purposes of this discussion, where
the context permits, an installed copy of the IBM LAN to LAN Wide Area Network
Program is referred to as a NetBIOS switch.
The IBM LAN to LAN Wide Area Network Program (a NetBIOS switch):
o Allows existing LAN-attached NetBIOS stations to communicate via a WAN.
o Is able to use an SNA backbone (SNA subarea or SNA Advanced Peer-to-Peer
(APPN)) network or an X.25 Packet Switched Data Network (PSDN) as the WAN.
o Is an application designed to run on a Personal Computer or Personal
System/2 machine under the OS/2 Extended Edition Version 1.2. operating
system.
o Has been designed to use the OS/2 Extended Edition Communication Manager
Advanced Peer-to-Peer Communication (APPC) LU6.2 interface to send and
receive traffic between NetBIOS switches.
o Supports one-to-many and many-to-one NetBIOS switch configurations.
o Provides a graphical and text based operator interface for planning,
configuration, installation, monitoring, error detection and recovery.
Figure 113 under heading "6.9.1.1.1 General Description" shows a simple
configuration.
-------------------------------------------------------------------------------
*------------*
] WAN ]
*-----* ] ] *-----*
]NBIOS] *---------* ] ] *---------* ]NBIOS]
]Stn. ]--] LAN ] ]--* *--] ] LAN ]--]Stn. ]
]A1 <......... ] ]GW] ]GW] ] .........> B1]
*-----* ] . ] ] ] ] ] ] . ] *-----*
] . ] ] ] ] ] ] . ]
*-----* ] . ########################## . " *"""""*
]A2 <......... # ] ] ] ] ] ] # .........> B2]
]NBIOS]--] .. # ] ]--* *--] ] # .. ]--]NBIOS]
]Stn. ] *--..---#-* ] ] *-#---..--* ]Stn. ]
*-----* .. ] # ] ] # ] .. *-----*
*..---#-* ] ] *#---..-*
].. # ] ] ] ]# .. ]
]VV V ] ] ] ]V VV ]
] ] *------------* ] ]
]LAN-LAN] ]LAN-LAN]
]WAN pgm] ]WAN pgm]
*-------* *-------*
switch A switch B
<------------> <------------>
Region A Region B
<###> LU 6.2 session
<...> 802.2 Connection
NBIOS Station running NetBIOS application
Stn.
-------------------------------------------------------------------------------
Figure 113. Connection of LANs via a WAN using IBM LAN to LAN Wide Area Network
Program
In this diagram, a NetBIOS application in station A1 is communicating with a
partner in station B1. Station A1 might be a DOS LAN Requester accessing a
file server in station B1. At the same time, station A2, perhaps a print
server, is being accessed by station B2. Both connections are multiplexed
onto an APPC LU 6.2 SNA session between the two NetBIOS switches. In fact,
there actually are two LU 6.2 sessions and conversations between the switches,
though only one is shown. One of these sessions is used for sending data
between switch A and switch B, the other for data travelling in the opposite
direction.
The diagram indicates that the LANs, with their respective NetBIOS switches
can be termed a REGION. A region is a self-contained group of LAN segments
connected together by MAC level bridges. Within a region, station MAC
addresses and segment numbers must be unique. When regions are connected
together with NetBIOS switches, segment numbers and MAC addresses may be
duplicated between regions. This is an important migrational benefit, since
when regions are to be connected for the first time, a major renumbering task
may not be required. If the regions were to be connected by a remote bridge,
then uniqueness would be a requirement. The regions would be merged into one.
The IBM LAN to LAN Wide Area Network Program supports multiple connectivity
options. On the LAN side the following are supported:
o IBM Token-Ring
o IBM PC Network (Broadband and Baseband)
o Ethernet.
On the WAN side, any communication method that can be used by OS/2 Extended
Edition 1.2 for APPC is supported. This includes all three LAN types described
above, as well as:
o Synchronous Data Link Control (SDLC), using the standard IBM PC or PS/2
line driver cards. This link could run directly to another NetBIOS
switch, to an SNA Communications Controller, to an IBM 9370 or an IBM
AS/400 processor.
o CCITT X.25 using the appropriate PC or PS/2 card and interfacing to a
public or private PSDN.
MANY-TO-ONE or ONE-TO-MANY
A NetBIOS switch is capable of having LU6.2 sessions with many other switches.
This means that when multiple regions are to be connected, only one copy of
the IBM LAN to LAN Wide Area Network Program is needed in any region. However,
if there are two switches in the same region, both cannot have sessions with
one or more switches in another region. If they did, then a "parallel path"
would exist between the two regions, and this is not supported. The reason for
this will become clear later. Reference should now be made to Figure 114
under heading "6.9.1.1.1 General Description".
-------------------------------------------------------------------------------
*-------*
]LAN-LAN]
.... . ]WAN pgm] ] ] ]
... . ] ]--+--+--+-
. . .. ]switch ] ] ] ]
*-------* . ....#####] A ] Region A
]LAN-LAN] . #... *-------*
]WAN Pgm] . # .
] ]############ WAN # .
]switch ] . # # .
] B ]#### ... ############### # .
*-------* # ..... # # .
-+- # ############# . #.#..
] # # ... # . # #
Region ] ### # . .# # #
B ] -+- # # . . # # # *-------*
--+--* *-------* . # # # ]LAN-LAN]
] ]LAN-LAN] # # ####]WAN pgm] *-------------*
]WAN pgm] # # ] ]---] Token-Ring ]
*----------* ] ] # ######] ] *-------------*
]Token-Ring]--]switch ] #########] D ] Region D
*----------* ] C ] *-------*
Region C *-------*
-------------------------------------------------------------------------------
Figure 114. One-to-Many Configuration
Figure 114 under heading "6.9.1.1.1 General Description" shows that a station
on the token-ring in region D is able to communicate with a station in any of
the other three regions, because the NetBIOS switch in region D has a session
with each of the other switches. However, no station in region B can
communicate with any station in region A, since there is no session between
NetBIOS switch B and NetBIOS switch A. Although there is a session between
switch B and switch D and another from D to A, a "concatenated" connection is
not supported.
6.9.1.1.2 The LU 6.2 Sessions: The purpose of the LU 6.2 sessions is to
provide a 'pipe' for communication between the NetBIOS switches. This pipe is
used for NetBIOS data transport between one LAN and the other, for setting up
the communications path between the NetBIOS applications, and for internal
communication between the switches for flow control and error information.
Between any two switches, there are two sessions, one for transmitting data,
the other for receiving data. The purpose of this implementation is to make
the data path between the switches full duplex.
The sessions are able to provide reliable data transport over the WAN, as well
as being able to use SNA pacing to provide a flow control mechanism. A
"pacing function", using the 802.2 logical link control is also provided
between the switches and the stations within their region. This ensures that
no part of the connection path can become over-congested with data.
The APPC sessions are configured using the facilities of OS/2 EE 1.2
Communication Manager. A single logical unit (LU) is defined, with the
capability for this LU to have LU-LU sessions with multiple partners.
The IBM LAN to LAN Wide Area Network Program is defined so as to have
knowledge of its own LU, its own region, and the LU names of its partners. The
region name of the partner is learned when the partner session is established.
The LU sessions can be permanent, started automatically when the program is
initialized, or temporary, that is started by operator command. If the partner
switch is not active when session initialization is attempted, and the session
is permanent, then the session will be retried. If the partner switch is not
active when session initialization is attempted, and the session is temporary,
the initiating switch will wait for the partner switch to initiate the
session.
The status of the sessions can be displayed from the operator's screen,
together with operational statistics.
6.9.1.1.3 WAN Connectivity: The following text in association with
Figure 115 under heading "6.9.1.1.4 Direct switch to switch Connection"
through Figure 119 under heading "6.9.1.1.5 Connection using a Backbone WAN"
shows some of the WAN connectivity options open to users of the NetBIOS
switch.
6.9.1.1.4 Direct switch to switch Connection: In this example, NetBIOS
switches are connected together with a single leased line.
-------------------------------------------------------------------------------
SDLC LINK
<-APPC--> <-APPC-->
*----------------------* Session *----------------------*
] ] ] --+---+---+-----------+---+---+-> ] ] ]
] N ] LAN ] ] ] ] LU 6.2 ] ] ] ] LAN ] N ]
] E ] to ] ] P ] S ] ] S ] P ] ] to ] E ]
L ] T ] LAN ] L ] U ] D ]-----/ ] D ] U ] L ] LAN ] T ] L
A<-] B ] ] U ] 2 ] L ] /-------] L ] 2 ] U ] ] B ]->A
N ] I ] WAN ] ] . ] C ] ] C ] . ] ] WAN ] I ] N
] O ] pgm. ] ] 1 ] ] ] ] 1 ] ] pgm. ] O ]
] S ] ] ] ] ] LU 6.2 ] ] ] ] ] S ]
] ] ] <-+---+---+-----------+---+---+-- ] ] ]
*----------------------* Session *----------------------*
-------------------------------------------------------------------------------
Figure 115. Direct Connection - SDLC
It is only possible to connect two NetBIOS switches together in this way,
since the SDLC link is ONLY supported as a point-to-point link under OS/2.
OS/2 Extended Edition can support up to two SDLC links, (by installing two
multiprotocol adapter cards), which means that if direct links are being used,
it is only possible for one switch to communicate with two others. The
simplest topology that can illustrate this is the triangle of three switches,
each with a link to the other two.
The speed of the SDLC link is limited to 19.2Kbps.
-------------------------------------------------------------------------------
X.25
*----------------------* Session *----------------------*
] ] ] --+---+---+-----------+---+---+-> ] ] ]
] N ] LAN ] ] ] X ] ] X ] ] ] LAN ] N ]
] E ] to ] ] P ] . ] ] . ] P ] ] to ] E ]
L ] T ] LAN ] L ] U ] 2 ]---X.25 ] 2 ] U ] L ] LAN ] T ] L
A<-] B ] ] U ] 2 ] 5 ] Network--] 5 ] 2 ] U ] ] B ]->A
N ] I ] WAN ] ] . ] ] ] ] . ] ] WAN ] I ] N
] O ] pgm. ] ] 1 ] V ] ] V ] 1 ] ] pgm. ] O ]
] S ] ] ] ] C ] ] C ] ] ] ] S ]
] ] ] <-+---+---+-----------+---+---+-- ] ] ]
*----------------------* Session *----------------------*
-------------------------------------------------------------------------------
Figure 116. Direct Connection - X.25
One X.25 virtual circuit is required for each switch-to-switch connection.
However, since these virtual circuits can be multiplexed onto a single leased
link to the X.25 network, one switch may simultaneously connect to many
others.
In order to use X.25, the SDLC adapter is replaced by an IBM X.25
Co-Processor/2 card for a Micro-Channel Architecture Personal System/2
machine. Link speed can be 64Kbps maximum.
6.9.1.1.5 Connection using a Backbone WAN: If using SDLC links to an IBM
Communications Controller with NCP, a scenario is illustrated in Figure 117
under heading "6.9.1.1.5 Connection using a Backbone WAN".
When using an SNA subarea network, many switches can be supported because of
the routing ability of the SNA backbone.
In the following diagrams the sessions between the NetBIOS switches have been
shown as one, for the sake of clarity. All sessions are of course routed
through the Communication Controllers containing the NCP.
In order to support the IBM LAN to LAN Wide Area Network Program and APPC as
INDEPENDENT LUS ACF/VTAM is required to be at Version 3.2 or above, with
ACF/NCP at level Version 5.2 or above for the IBM 3745 and 3720 controllers,
and Version 4.3 for the IBM 3725.
-------------------------------------------------------------------------------
USING IBM COMMUNICATIONS CONTROLLERS
LAN LAN
A A
] ]
*----------------* *----------------*
] NetBIOS ] ] NetBIOS ]
]----------------] ]----------------]
]LAN-LAN WAN pgm.] ]LAN-LAN WAN pgm.]
]----------------] ]----------------]
]A LU AA] ]AA LU A]
]+-------------++] ]++-------------+]
]] PU 2.1 ]]] ]]] PU 2.1 ]]
]+-------------++] ]++-------------+]
]] SDLC or LAN ]]] ]]] SDLC or LAN ]]
*+-------------++* *++-------------+*
] ] ]*----------------*] ] ]
----* ] ] *------------------+-----+------*]
] ] *---------* ]*---------* ]]
V ] ] ] ]-----------/ ]] ] ]]
T ]- - -]-] NCP ] /---------+] NCP ] ]]
A ] ] ] ] ]] ] ]]
M ] ] *---------* *------------------**---------* ]]
] ] ] ]*----------------* ] ]]
----* ] ] ]] ] ] ]]
*+-------------++* *+-------------++*
]] SDLC or LAN ]]] ]] SDLC or LAN ]]]
]+-------------++] ]+-------------++]
]] PU 2.1 ]]] ]] PU 2.1 ]]]
]+-------------++] ]+-------------++]
]V LU VV] ]V LU VV]
]----------------] ]----------------]
]LAN-LAN WAN pgm.] ]LAN-LAN WAN pgm.]
]----------------] ]----------------]
] NetBIOS ] ] NetBIOS ]
*----------------* *----------------*
] LU 6.2 ]
V <--------> V
LAN Sessions LAN
-------------------------------------------------------------------------------
Figure 117. SNA Subarea Network
The NCPs could be replaced by an IBM 9370, its Integrated Communication
Adapter and VM/9370 Version 3.3
Almost identical pictures can be drawn for connection across an X.25 network.
The 37xx/NCP must now contain NCP Packet Switching Interface (NPSI), and the
SDLC link in the NetBIOS switches is replaced by an X.25 virtual circuit.
NPSI must be at a level compatible with the NCP installed.
In this example, the SNA subarea network is accessed via an X.25 network.
Routing for the APPC sessions, is still dependent on the capabilities of the
SNA subareas. The scenario is illustrated in Figure 118 under heading
"6.9.1.1.5 Connection using a Backbone WAN".
-------------------------------------------------------------------------------
SNA OVER X.25
LAN LAN
A A
] ]
*----------------* *----------------*
] NetBIOS ] ] NetBIOS ]
]----------------] ]----------------]
]LAN-LAN WAN pgm.] ]LAN-LAN WAN pgm.]
]----------------] ]----------------]
]A LU AA] ]AA LU A]
]+-------------++] ]++-------------+]
]] PU 2.1 ]]] ]]] PU 2.1 ]]
]+-------------++] ]++-------------+]
]] X.25 VC ]]] ]]] X.25 VC ]]
*+-------------++* *++-------------+*
] ] ]] ]] ] ]
] X.25 ]*----------------*] X.25 ]
----* ] ] *------------------+-----+------*]
] ] *---------* ]*---------* ]]
V ] ] ] NCP ] ]] NCP ] ]]
T ]- - -]-] NPSI ]-----X.25 Network---+] NPSI ] ]]
A ] ] ] ] ]] ] ]]
M ] ] *---------* *------------------**---------* ]]
] ] ] ] ] ]]
----* ] X.25 ]*----------------* X.25 ]]
] ] ]] ] ] ]]
*+-------------++* *+-------------++*
]] X.25 VC ]]] ]] X.25 VC ]]]
]+-------------++] ]+-------------++]
]] PU 2.1 ]]] ]] PU 2.1 ]]]
]+-------------++] ]+-------------++]
]V LU VV] ]V LU VV]
]----------------] ]----------------]
]LAN-LAN WAN pgm.] ]LAN-LAN WAN pgm.]
]----------------] ]----------------]
] NetBIOS ] ] NetBIOS ]
*----------------* *----------------*
] ]
V V
LAN LAN
-------------------------------------------------------------------------------
Figure 118. SNA Subarea Network over X.25
Another alternative is to use an SNA Advanced Peer-to-Peer network. This is
currently implemented with the IBM AS/400 system and is shown in Figure 119
under heading "6.9.1.1.5 Connection using a Backbone WAN".
-------------------------------------------------------------------------------
AS/400 APPN
LAN LAN
A A
] ]
*----------------* *----------------*
] NetBIOS ] ] NetBIOS ]
]----------------] ]----------------]
]LAN-LAN WAN pgm.] ]LAN-LAN WAN pgm.]
]----------------] ]----------------]
]A LU AA] ]AA LU A]
]+-------------++] ]++-------------+]
]] PU 2.1 ]]] ]]] PU 2.1 ]]
]+-------------++] ]++-------------+]
]] SDLC or LAN ]]] ]]] SDLC or LAN ]]
*+-------------++* *++-------------+*
] ] ]*----------------*] ] ]
] ] *------------------+-----+------*]
] *---------* ]*---------* ]]
] ] ]----------/ ]] ] ]]
] ] AS/400 ] /----------+] AS/400 ] ]]
] ] ] ]] ] ]]
] *---------* *------------------**---------* ]]
] ] ]*----------------* ] ]]
] ] ]] ] ] ]]
*+-------------++* *+-------------++*
]] SDLC or LAN ]]] ]] SDLC or LAN ]]]
]+-------------++] ]+-------------++]
]] PU 2.1 ]]] ]] PU 2.1 ]]]
]+-------------++] ]+-------------++]
]V LU VV] ]V LU VV]
]----------------] ]----------------]
]LAN-LAN WAN pgm.] ]LAN-LAN WAN pgm.]
]----------------] ]----------------]
] NetBIOS ] ] NetBIOS ]
*----------------* *----------------*
] ]
V V
LAN LAN
-------------------------------------------------------------------------------
Figure 119. AS/400 Advanced Peer-to-Peer Network
6.9.1.1.6 Determining the Target Region: As has been discussed in "5.6.1.2
NetBIOS Interface" NetBIOS uses NAMES rather than MAC addresses for session
support, and provides a set of services to locate and associate these names
with network MAC addresses. A NetBIOS requester, perhaps named PETE, wants to
access resources located on a NetBIOS server with the name SRV. When PETE
wants to establish a session with SRV, a command will be sent on the network
to find which station knows the name SRV. Once located, a session will be
started, and data transferred. Names must therefore be unique.
The command used for this initial search function is the NetBIOS NAME_QUERY ,
and it is addressed to the NETBIOS FUNCTIONAL ADDRESS. Because the IBM LAN to
LAN Wide Area Network Program has this address opened on it's LAN adapter, it
is able to receive the NetBIOS Name_Query command.
The NetBIOS switch then searches its user-defined FILTER AND TARGET RESOLUTION
TABLE to determine if it needs to route the request, and if so to which
region. The table has two purposes which are performed sequentially:
1. RESOLUTION Determine the target region for a given name. If the name is a
NetBIOS GROUP NAME, then there may be a need to route the request to
multiple regions.
2. FILTERING Given the target region(s) and name, determine whether the
frame should be forwarded to the region or not.
NAME RESOLUTION
At each IBM LAN to LAN Wide Area Network Program, the user must configure a
list of all the NetBIOS names on that region which are to be accessed by other
regions. To keep the user from having to determine the actual 16 character
NetBIOS name (which may be significantly different than the application name),
the names in the list have implicit wild cards at the beginning and end of the
name. For example, if the name PETE was entered in the list, destination name
'PETER ' and 'DOMAINxxPETE ' would find a match.
Again, the list is configured at the IBM LAN to LAN Wide Area Network Program
on which the names reside. When the IBM LAN to LAN Wide Area Network
Program-to-IBM LAN to LAN Wide Area Network Program sessions are established,
the list is passed to the partner IBM LAN to LAN Wide Area Network Program.
When a NetBIOS NAME_QUERY or DATAGRAM frame is received at a IBM LAN to LAN
Wide Area Network Program the destination name is matched against every name
in every target region's list until a match is found or the list is exhausted.
If a match is found, the frame is passed onto the target region. Otherwise it
is discarded. The frame is NOT broadcast to all regions in search of the
name.
If only servers are to be accessed in a region, the name list could consist of
each server name or just the domain name of a group of servers. If
peer-to-peer communications was desired, it may be necessary to rename all
stations in a region to imbed a region qualifier in the name. This region
qualifier would then be added to the name list.
FILTERING
The filter function gets control on each broadcast frame to a specific name or
group name. Its major function will be, given the target region for a frame,
to determine whether the frame should be forwarded to that region or filtered.
The filter routine will be a dynamic link routine external to the rest of the
IBM LAN to LAN Wide Area Network Program product. The interface to the filter
will be published so that a customer can write and install his own filter
routine tailored to the particular network needs.
In addition, the IBM LAN to LAN Wide Area Network Program will provide on the
shipped diskette a general purpose filter routine.
6.9.1.1.7 NetBIOS Session Setup
-------------------------------------------------------------------------------
*------* *------* *------* *------*
] PETE ] ]switch]----------->]switch] ] SRV ]
] ]<-->] ] WAN ] ]<-->] ]
] ]LAN ] ]<-----------] ] LAN] ]
*------* *------* *------* *------*
Name_Query -------->
(Step 1 under heading "6.9.1.1.7 NetBIOS Session Setup")
Table (Step 2 under heading "6.9.1.1.7 NetBIOS Session Setup")
Search
------------------->
Name-Query (Step 3 under heading "6.9.1.1.7 NetBIOS Sess
------------>
(Step 4 under heading "6.9.1.1.7 NetB
<------------
<-------------------
<------------- Name_Recognized
(Step 5 under heading "6.9.1.1.7 NetBIOS Session Set
LLC
Connect------------>
LLC (Step 6 under heading "6.9.1.1.7 NetB
Connect------------>
<------------
(Step 7 under heading "6.9.1.1.7 NetBIOS Session Setup")
<------------
-------------------------------------------->
Session_Initialize
(Step 8 under heading "6.9.1.1.7 NetBIOS Session Setup")
<--------------------------------------------
Session_Confirm
<------------------------------------------->
Data Flow
*------* *------* *------* *------*
] PETE ] ]switch]----------->]switch] ] SRV ]
] ]<-->] ] WAN ] ]<-->] ]
] ]LAN ] ]<-----------] ] LAN] ]
*------* *------* *------* *------*
]<----------->] ]<------------->] ]<----------->]
802.2 Circuit LU6.2 session 802.2 Circuit
]<---------------------------------------------->]
NetBIOS SESSION
-------------------------------------------------------------------------------
Figure 120. NetBIOS Session Initiation Flow across a WAN
We will now describe what happens when a cross-WAN NetBIOS session is to be
setup. Reference can be made to Figure 120 under heading "6.9.1.1.7 NetBIOS
Session Setup" which shows the flows, together with a pointer to the
descriptive text.
1. When the NetBIOS Requester with the name PETE wishes to establish a
session with the NetBIOS server SRV, PETE's NetBIOS application will send
a NAME_QUERY command as a Single-Route Broadcast frame addressed to the
NetBIOS functional address. The frame will include the name SRV, with the
expectation that a station will recognize the name and respond.
2. Since IBM LAN to LAN Wide Area Network Program has opened the NetBIOS
functional address it will receive commands sent to the NetBIOS functional
address, and search the resolution table to determine if the name is
defined, and in what region it lies. Determining the region that a
NetBIOS name is associated with is done on an LU 6.2 session basis. So if
one IBM LAN to LAN Wide Area Network Program is connected to several other
IBM LAN to LAN Wide Area Network Programs, as that one IBM LAN to LAN Wide
Area Network Program connects to another IBM LAN to LAN Wide Area Network
Program the NetBIOS names that are forwarded would be associated with that
particular LU 6.2 session. That one IBM LAN to LAN Wide Area Network
Program would have a Name Resolution table that would associate any given
NetBIOS name to the LU 6.2 session that passed the name to it.
Definitions at a local IBM LAN to LAN Wide Area Network Program would only
be for those local NetBIOS names that would want to have access through
the local IBM LAN to LAN Wide Area Network Program. It is not necessary
that any target NetBIOS names be known to the local IBM LAN to LAN Wide
Area Network Program.
3. Assuming that the target NetBIOS name is found, and the request is not
filtered out, the SOURCE NetBIOS switch will forward the Name_Query to
the TARGET switch over the appropriate pre-established LU 6.2 session.
4. The target switch then places the Name_Query message onto it's local LAN.
5. Assuming that SRV is active, a NAME_RECOGNIZED response will be returned
via the switches to the originating station.
6. Reception of the Name_Recognized at the target switch causes it to
initiate a connection-oriented logical link between itself and the target
station, SRV.
7. The NetBIOS switch (source switch) in PETE's region passes the
Name_Recognized back to PETE, giving the MAC address of SRV as the THE MAC
ADDRESS OF THE SOURCE NETBIOS SWITCH, not the actual MAC address of SRV.
PETE now starts a logical link between its own station and the source
switch or local switch.
This explains the reason why parallel NetBIOS switch paths are not
supported between regions. If a parallel path existed, there would be two
or more source switches. Both would forward the Name_Query to the target
region, and both sources would provide responses. In effect, it would look
as though SRV existed on the network twice, at two different MAC
addresses. This condition is a violation of NetBIOS protocol.
Note that the initiation of steps 6 and 7 occur simultaneously.
8. Once the logical link connection at the source end is completed a
SESSION_INITIALIZE is sent by PETE to SRV, and the SESSION_CONFIRM
response indicates that data transfer can begin.
The session is now established. The NetBIOS session flows end-to-end, between
PETE and SRV. However, two logical links have been established, one between
PETE and the source switch, the other between SRV and the target switch. Since
the performance characteristics of the WAN are unknown, there is no
implication on the logical links for timing problems or delays within the WAN,
though adjustment of NetBIOS session timers may be required.
One 802.2 logical link and several NetBIOS sessions from any one workstation
are known as one 802.2 CIRCUITs. An 802.2 Circuit is an internal resource to
the IBM LAN to LAN Wide Area Network Program. There are a finite number of
802.2 Circuits available. When a station wants to establish a NetBIOS session
with another station using the IBM LAN to LAN Wide Area Network Program, an
802.2 logical link is established between the station and the local IBM LAN to
LAN Wide Area Network Program. A NetBIOS session is also established between
each station. One 802.2 Circuit is then allocated from the IBM LAN to LAN
Wide Area Network Program workstation. If the same station was to establish
another NetBIOS session, it would use the one 802.2 Circuit that had already
been established. This permits the IBM LAN to LAN Wide Area Network Program
to minimize the use of 802.2 link stations and allow more NetBIOS sessions
across the WAN.
6.9.1.1.8 Management
6.9.1.1.9 Alert Management: The IBM LAN to LAN Wide Area Network Program
sends its alerts to the LAN Manager using the Alert Transport Facility of IBM
LAN Manager V2 and IBM LAN Manager Entry or above. Since this facility does
not use the NetBIOS protocol, a LAN Manager application must be present in
each LAN region. This LAN Manager may of course be connected to NetView on a
S/370 host for central management.
Alerts are provided for:
o The defined number of 802.2 Circuits has been exceeded
o No more 802.2 link stations available
o Internal buffers depleted
o LU 6.2 session dropped
o IBM LAN to LAN Wide Area Network Program terminated.
6.9.1.1.10 Operator Interface: The operator of the IBM LAN to LAN Wide Area
Network Program can modify some dynamic parameters without re-initializing the
program, and obtain information about the following:
o The OS/2 EE Communication Manager configuration
o Fixed parameters of the switch configuration
Includes LU Names, buffer sizes and numbers, LAN adapters used, and filter
used.
o Dynamic configuration parameters
Aging timers, warning thresholds, timing intervals for report logging
logging and resolution table are examples.
o Problem determination and performance/utilization
A range of counters, thresholds, status indicators and time/date
information on:
- The switch itself
- LU 6.2 sessions
- NetBIOS sessions
- 802.2 Circuits
- Errors
o Viewing the disk log
A disk based log is an option of the IBM LAN to LAN Wide Area Network
Program. Information may be placed in this file as a result of specific
events, or at regular intervals.
Facilities such as the OS/2 EE Communications Manager trace and message files
are also available to assist with problem determination.
6.9.1.1.11 Limitations: The IBM LAN to LAN Wide Area Network Program has the
following limitations:
o Each IBM LAN to LAN Wide Area Network Program cannot be configured for
connection to more than 48 remote IBM LAN to LAN Wide Area Network
Programs.
o The IBM LAN to LAN Wide Area Network Program must reside within the same
workstation as the OS/2 EE V1.2 SNA gateway.
o Duplicate NetBIOS names should not be defined to the IBM LAN to LAN Wide
Area Network Program or reside within the network. A warning message will
be issued.
6.9.1.1.12 Summary: The IBM LAN to LAN Wide Area Network Program:
o Provides a means for interconnecting LANs across a wide area network to
convey NetBIOS traffic from one LAN to another.
The WAN can be a SNA subarea network, an SNA APPN network, or an X.25
network. An X.25 network must be capable of supporting Qualified Logical
Link Control (QLLC - SNA over X.25).switches can be connected together by
a dedicated link.
o Allows the NetBIOS traffic to share the bandwidth of the WAN with other
traffic.
o Uses APPC LU6.2 sessions as "pipes" for the inter-regional NetBIOS
traffic.
o Uses the facilities of OS/2 Extended Edition Communications Manager 1.2 to
support the APPC LU 6.2 communications.
o Can attach to Ethernet, token-ring or PC Network (Broadband and Baseband)
LANs.
o Is an OS/2 application, therefore need not run in a dedicated machine.
NOTE: If the machine is required to run another NetBIOS application
apart from the IBM LAN to LAN Wide Area Network Program, the LAN adapter
cannot be shared between the applications. Two adapters must be installed.
o Provides a region resolution and filter mechanism to control access to
resources across the WAN. The filter is tailorable to customers' needs.
o Does not support "concatenated" or "parallel" connections between regions.
o Is provided with a set of utilities, allowing operator control, statistics
gathering and comprehensive problem determination facilities.
o Is positioned for customers who have made an investment in their WANs and
who do not wish to invest in the link costs of a remote bridging solution.
The product should be used by customers who regard their inter-LAN NetBIOS
traffic as a small percentage of their overall WAN traffic. It is very
suitable for those customers that only want to provide occasional access
to NetBIOS resources through the WAN.