Brought to you by EarthWeb
IT Library Logo

Click Here!
Click Here!


Search the site:
 
EXPERT SEARCH -----
Programming Languages
Databases
Security
Web Services
Network Services
Middleware
Components
Operating Systems
User Interfaces
Groupware & Collaboration
Content Management
Productivity Applications
Hardware
Fun & Games

EarthWeb Direct EarthWeb Direct Fatbrain Auctions Support Source Answers

EarthWeb sites
Crossnodes
Datamation
Developer.com
DICE
EarthWeb.com
EarthWeb Direct
ERP Hub
Gamelan
GoCertify.com
HTMLGoodies
Intranet Journal
IT Knowledge
IT Library
JavaGoodies
JARS
JavaScripts.com
open source IT
RoadCoders
Y2K Info

Previous Table of Contents Next


MESSAGE STRUCTURE DEFINITION

In addition, Fibre Channel defines message structure to allow disassembly and reassembly of messages from the application layer of the transmitting side to the same level at the receiving side. These are called Fibre Channel exchanges, sequences, and frames.

Exchanges

An exchange refers to an application layer functional communication between users. Examples are a SCSI connection or a TCP/IP session. Several exchanges can be active simultaneously.

Sequences

In Fibre Channel lexicon, a sequence is a related set of frames (one or more) within the context of a specific exchange. Sometimes a sequence is called the Fibre Channel information unit (though not specifically defined by the standard).

Frame

A frame is a set of bytes containing control (i.e., routing) and data elements. A frame is the atomic unit in Fibre Channel. Frame overhead is a fixed 36 bytes. Frames are variable from 36 bytes (control information only) to 2,112 bytes long, which allows it to be very efficient for normal-size exchanges between computers as well as to and from mass storage connections. Exhibit 2-4-2 illustrates framing, message assembly, and message disassembly.


Exhibit 2-4-2.  Fibre Channel Information Transfer

Fibre Channel further defines classes of messages that encompass both circuit connection (class 1) and packet connection (classes 2, 3, 4). Class 2 guarantees delivery and acknowledgment, a unique feature in packetized protocol technologies that improves transfer efficiency because:

  The receiver does not need to guess about delivery.
  Packets cannot be dropped without notification to the sender; the receiver sends back an acknowledgment indicating that the data has been received and whether or not it has been corrupted (i.e., a CRC check).

Many small-scale implementations use class 3 (i.e., packetized datagram- unreliable) because it is simpler to implement and relatively reliable.

SCALABILITY AND GUIDELINES FOR USE

Forms of Fibre Channel connection include point-to-point, arbitrated loop, and switched fabric (see Exhibit 2-4-3). Users should select the connection that fits their application without device-cost impact. Topologies are designed to be interconnectable and upgradable.


Exhibit 2-4-3.  Fibre Channel Topologies

Point-to-Point Connections

Point-to-point links provide a dedicated high-bandwidth connection, such as the one from a disk array to a server. Some forms provide high availability, fault tolerant access for two-server/two-disk-array applications.

Fibre Channel Fabrics

In the switched fabric topology, traffic between Fibre Channel ports passes through an intermediary switch called a fabric that combines cross-bar and packet-switching capabilities. Multiple switches can be linked together to form larger fabrics.

Like other switched media (e.g. Fast Ethernet switching hubs or asynchronous transfer mode [ATM] switches), capacity in switched fabric topologies is scalable. New switch modules are added to connect additional devices and the aggregate throughput increases to accommodate the increased load.

Fabrics can also interconnect devices with different speed and Fibre Channel media, so the speed and cost of the Fibre Channel link can be selected to match the anticipated traffic from the bridge or router.

Fibre Channel fabrics provide multicast services without requiring external frame replication servers, although the size of a multicast packet is limited to the maximum frame size of the fabric (generally 2K bytes). Available fabrics support datagram, acknowledged connectionless, and connection-oriented service classes. In the future, some fabrics will support a virtual circuit-oriented class of service, in which multiple circuits with guaranteed bandwidth can be established by a single port.

Arbitrated Loop

In the arbitrated loop topology, up to 126 nodes can be connected in a shared-media topology. Instead of a circulating token or a collision sense mechanism, a simple arbitration mechanism ensures fairness. Star topologies with passive hubs provide robustness; some hub ports may be equipped with a loop bypass circuit to ensure continued loop operation if the port is offline.

Current implementations use TwinAxcess operating at a gigabit, which has ample capacity for interconnecting disk drives that can source on the average less than 20M bytes or network connections with 10M bps and a certain number of 100M-bps LANs (Ethernet).

Low cost is the primary advantage of the arbitrated loop topology. Only half the number of transceivers are required compared with fabric connections, and no switch is required. However, because bandwidth is shared among all nodes on an individual loop, the number of devices will be limited if many have high-traffic requirements.

Fibre Channel Application Interconnection

Fibre Channel can be viewed as an application-to-application connection using a common adapter port, such as a workstation, PC, storage array, disk drive, or any other appropriate peripheral (this port is known as the N_Port in Fibre Channel lexicon) and a common interconnection technology (i.e., point-to-point, loop, or switched). Exhibit 2-4-4 illustrates simultaneous multiplexed connections among applications using Fibre Channel.


Exhibit 2-4-4.  Simultaneous Multiplexed Connections

Many applications can benefit from the economies of scale of other applications, with the incremental costs being small for both the hardware and software elements. For little additional complexity, a connection can be both serial SCSI and TCP/IP. Examples include multiple video connections (being investigated by digital video and digital movie studios), clustering and disk file access, and interactive medical applications that require streaming of data from an imaging device (e.g., MRI) and simultaneously sending the data via TCP/IP to a doctor's office in the same facility.

SUMMARY

Because of its unique architecture, Fibre Channel can offer the high-performance, low-cost connection required by bandwidth-intensive applications. By combining the attributes of a channel and a network, Fibre Channel enables data transfer rates that are up to 250 times faster than many network protocols. Now that computers are faster and better able to handle large amounts of data, a network interconnection is needed that can handle higher speeds. Fibre Channel is emerging as the information connection technology that enables reliability and performance at record speeds.


Previous Table of Contents Next

footer nav
Use of this site is subject certain Terms & Conditions.
Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details.