Internet Standards/CD1.mdf/MISC/MERIT/NOOP/NOTES/11_20_91.AM

Text File | 1992-03-06 | 11.6 KB | 388 lines

                   Network OSI Operational Working Group
                                 Minutes
                             November 20, 1991
                           Santa Fe, New Mexico

Chair:
     Sue Hares/Merit ([e-mail])

Mailing List:
     [e-mail]  (administrative requests only)
     [e-mail]

Minutes:
     These minutes cover the Wednesday evening session held from
     9:30am - 12:00am.

Next meeting: March IETF meeting (probably Monday)

Note Taker: April Merrill, NCSC
Editor: Sue Hares
(note the Editor heavily edited several sections, and encourages
people to comment on these notes to [e-mail].  These
minutes will not be considered final until the next IETF meeting.)

                           +-----------+
                           | Objective |
                           +-----------+

The objective of the session was discussed during the earlier meeting.
The notes do not relfect the actual topics.  The following topics
were dicussed:

1) Introductions and OSI at your site
2) OSI Interop 91
3) Changing the Charter
4) Regional Routing Plans
5) NSFNET T1 OSI Routing Plan
6) NSFNET T3 OSI Routing Plan
7) DECNET Phase 4 -> Phase 5 problems
8) Regional Routing Discussion
  - transition to production OSI service
  - National Test Beds for OSI
  - Survey for OSI service
  - Security Concerns
9) OSI tools
10) NSAP addressing
11) OSI HitchHickers Guide


Discussion topics:

1.) Introductions and OSI at your site


People attending the meeting gave their name and:

a) what is their interest in OSI
b) what is happening at their site for OSI


2.) OSI Interop '91

    IS-IS for 4 vendor 3COM, DEC,  proteon, wellfleet tested IS-IS
    interoperability by Interop '91. 3COM, proteon, and wellfleet
    provide IS-IS support for Interop '91 OSI Demonstration booth.

    IDRP prototype developed for Interop '91 by Dave Katz of MERIT.
     MERIT expects work on prototype to continue.  MITRE is
    also working on a prototype for IDRP.

    25 network service providers and 25 OSI vendors
    participated to provide Interop '91 connectivity for
    4 OSI applications: FTAM, VT, X.400, and X.500.


3.) Changing the Charter

    Sue Hares (working group chair) wants to change the charter.
    The group felt their was signifigant work to still
    be accomplished.  Sue Hares will re-write the charter
    and send it out the working group.  This charter should
    include:
        1.) Survey form for OSI service
        2.) RFC on OSI tools (late addition by Sue Hares
                     working group chair)
        3.) Tools work re-defined
        4.) FYI RFC on OSI in the Internet -
             Frequently asked questions
        5.) Pilot project activities
        6.) National Test bed activities
        7.) Routing Plans still written and reviewed



Action item 1: Re-write charter to update to current work
person: Sue Hares


4) Routing Plans

The group had a lot of concerns about how to make CLNS software work.
The routing software is not being well tested for CLNS and IP together.
Problems show up immediately or after software has been running for
a week or two weeks.

The Routing Plans are good tools.  However, their are 3 types of
routing plans:

a.) initial test routing plan
b.) medium term routing plan
c.) long term routing plan

Most people working for Interop '91 had initial test routing plans.
Most regional networks are going back and adapting medium term
routing plans.  This medium routing plan needs to try to look
at the long range routing ideas, but it needs to try to work something
out for now.

Sue Hares mentioned that any type of routing plan
is OK to send to the list.    Sue Hares was concerned that the
superb routing plans already sent to the noop people
have stopped people from sending initial attempts at routing plans
to the noop list.  Any thoughts or initial versions of routing plans
are welcomed on this list.

Additional regional routing plan discuss was tabled until
after the NSFNET T1 and T3 OSI routing plan  is described by
Sue Hares.

5) NSFNET T1 OSI Routing Plan

Architecture of T1 NSS: 9 RTs on dual token ring
History of OSI in NSFNET T1 Network:
    By Interop '89 - prototype demonstration
    By August '90 - full deploymnet on T1 network
    Interop '91 - used as part of OSI Internet demonstration

How OSI routing works:
    MAP NSAP prefix -> IP address
    then
    MAP IP address -> AS
    then
    MAP AS -> next hop

    When node reaches edge of NSFNET (external interface
    of remote NSS) then

    NSAP Prefix is mapped to Network Entity Title (NET) of
        router packets are to be sent to a regional network

    Both the NSAP prefix -> IP address and NSAP Prefix-> NET
        are static mappings.  Each nss processor (rcp and psp)
    has the NSAP prefix -> IP address mapping.   Only
    the E-PSP for which is attached to the NET has the mapping
    between the NSAP Prefix-> NET.

NSAP addresses
----------------
NSFNET uses GOSIP format for NSAP address assignment.

NSFNET NSAP has GOSIP IDP (47 0005), AAI of FFFF00.  The
Notes have the full NSAP address.

OSI routing strategy
---------------------
Transit all OSI traffic that conforms to
acceptable use strategy

6.) Proposed T3 OSI routing Architecture

Time for implementation:  as soon as possible to accomplish
the following things:

1) T3 network stability and transition from T1 network
2) Change some of the T3 Hardware to hardware that
    allows higher throughput
3) Change to software base to switch software to gated

Current plan is that the NSFNET T1 network will remain for OSI until
the NSFNET T3 has OSI working.

Proposed Software Support:

OSI support:
- CLNP support
- ES-IS support
- IDRP support in gated

Dual IS-IS in gated

(further details in the notes from the talk)

7) DECNET Phase 4 -> Phase 5 problems

DEC is shipping Phase 5 with VAX/vmx and Ultrix.
Tony Hain recommended bringing up Ultrix as phase 4.
Don't try Phase 5 yet on Ultrix if you already have
a Phase 4.   The DECNET Program for address assignment has
lots of power,  but may assign addresses you do not want
if the User does not understands the addressing questions.
DEC is working on additional refinements.

ESNET, DEC and Cisco are working on plans to
solve transition problems in transitioning between large DECNET
phase 4  areas (like HEPNET) to Phase 5 areas.

The ESNET routing plan will be out in January and has some
details on transition between phase 4 and phase 5.


8) Next step in CLNS routing Regional

OSI Infrastructure set-up for Interop '91 was  for  the
OSI demonstration.  The European WG-CLNS-4 has been running
both IP and CLNP since 1990.

John Curran shared about NEARNET's network.  The NEARNET
client sites are close together and have a high probability of
getting calls from DECNET Phase 5 sites.  NEARNET would like
advice.  Not enough people in regionals have worked in  this
area.  Sites can change from one concentrating router to
another hub router in the NEARNET backbone.  This switching
is possible due to the small distance between sites.  However,
it complicates the OSI routing.

John wonders if there is  way to assign them a long
term NSAP prefix.   One can assign NSAP prefixes for them to
get somethinging running for DECNET Phase 5, but NEARNET
is in the process of draawing up a plan for the NEARNET network
for OSI.  NEARNET expects to have a routing plan by December.

Trying to NSAP allocation in the NEARNET network is not clear
since their environement is changing.  John would like to see a few
OSI Networks operational before he decides how NEARNET will
handle OSI.

John defined some problems (with discuss from the group) we
face in OSI::

 - Customers was networks, but the OSI applications are not
   being used

 - Where do we get payback on the investiment in learning and
   technology for OSI service

 - Who will educate the people within the networks on OSI

 - We need to know which regionals are doing CLNS even if
   it is a partial CLNS service

 NOOP should do a survey of regional networks.

John Curran also defined 4 stages in getting CLNS from testing
to production:

Stage 1 - Initial experimentation

    Spare routers and test machines are used to
    try out the CLNS network layer and OSI applications.

Stage 2 - CLNS trial

    Need a national test bed to try out CLNS code on
    different routers.  These routers need to be tested
    under applications loads.  The Testbed could
    be glued together with IP networks using encapsulation.

Note the group decided to query the NOOP working group to
see who is interested in getting a Testbed.

Stage 3 - CLNS and IP in production network

    CLNS needs to run a test service in production routers.
However, the CLNS service is considered a "trial" service and
may encounter some down time.

(editor's note:  This test service in prodution routers was taken by
NSFNET in the T1 network, and has proven very effective.  The IP
service is given priority in problem solving, but the CLNS is
excersized)

Stage 4 - CLNS and IP production in network

    Both CLNS and IP are production services.


Users are reluctant to migrate to OSI due to need to cut
over applications or work with new OSI applications.  It is important
to get those OSI applications running in the Internet to run over
CLNP.  The X.500 and X.400 working groups in IETF should be encouraged
to get their applications running over TP4 and CLNP as well as
TCP/IP.  Also, we hope to see applications like X-windows transition
to OSI.


- National Test Bed

The NOOP group wants to organize a national test bed for
CLNP and applications over CLNP.

Action item 2:  Sue Hares will query NOOP and other lists to
    see who wants to start a working sub-group on a testbed
    for CLNP testing, and Phase 4 to Phase 5 transitions.



- Survey for OSI service

The NOOP group wants to have a surve of who will provide CLNP service.
Sue Hares, Linda Winkler, and John Curran will put together a list
of questions.

Action item 3: Put together a survey on OSI services
people: Sue Hares, Link Winkler, and John Curran


- Security Concerns

The NOOP noticed that none of the routers can filter packets based on
OSI addresses or OSI application information.  Companies which use
IP filter to provide some security for their company networks will
not let OSI traffic in from the Internet due to the lack of security
filters.   Sometimes OSI packets can flow to the router at a company, but
no further due to the lack of OSI filtering.



action item 4: Walt Lazear will write up the security concerns.

9) OSI tools

Tonight at 6:00pm - 7:00pm, a NOOP session will be helpd to discuss
network tools.  We need to start making recommendations on what
OSI network layer tools need to be in routers.

10.) NSAP addresses

Applications addresses will be in:

a.) X.500
b.) a flat file by sue
c.) Visual representation of OSI -> IP mapping
    for quick look-up in OSI debugging

d.) Domain Name Service

Action Item 6: Sue Hares will collect addresses and
    publish a list of file names for the flat files
    and their anonymous FTP location
person: Sue Hares

11.) Hitchickers guide to OSI in the Internet

Sue Hares is writting an FYI on OSI in the Internet.
Sue Hares will circulate this to the NOOP working group.





Summary of Action Items:

Action item 1: Re-write charter to update to current work
person: Sue Hares

Action item 2:  Sue Hares will query NOOP and other lists to
    see who wants to start a working sub-group on a testbed
    for CLNP testing, and Phase 4 to Phase 5 transitions.


Action item 3: Put together a survey on OSI services
people: Sue Hares, Link Winkler, and John Curran

Action item 4: Walt Lazear will write up the security concerns.
People: Walt Lazear

Action Item 5: Sue Hares will collect addresses and
    publish a list of file names for the flat files
    and their anonymous FTP location
person: Sue Hares

Action Item 6: Sue Hares will write and circulate for comment
    an FYI or Hitchhicker's guide to OSI in the Internet