Editor's note:  These minutes have not been edited.
 
Internetworking Over NBMA (ION) WG

The meeting minutes were taken by Karen O'Donoghue and Mike Davison,
and edited by Andy Malis.

Andy Malis and George Swallow chaired the working group meeting, which
met over two sessions, Monday 12/9/96 1530-1730, and Tuesday 12/10/96
1530-1730.  There were 276 attendees.

Andy presented the agenda and the current status of those WG documents
not being presented.  There was one new RFC since the Montreal
meeting: RFC 2022, Support for Multicast over UNI 3.0/3.1 based ATM
Networks, by Grenville Armitage. Congratulations and appreciation were
extended to Grenville.

The following drafts have completed WG last call:
draft-ietf-iplpdn-frmib-dte (in IESG review)
draft-ietf-rolc-nhrp (Submitted to IESG on 10/22)
draft-armitage-ion-cluster-size (Completed on 11/22)
draft-ietf-ion-bcast (11/22)
draft-ietf-ion-fr-update (12/6)
The last three will be submitted to the IESG in short order.

The following other drafts are in progress on the list:
draft-ietf-ion-mib, draft-ietf-ion-nhrp-mib, draft-ietf-ion-mars-mib,
draft-ietf-ion-sig-uni4.0, and draft-ietf-ion-inarp-update.

George gave a brief ATM Forum MPOA update. A major change was made to
the MPOA architecture that significantly reduced the complexity of the
effort.  A simpler client/server model has been developed, and SCSP
will be used to synchronize MPOA servers.  The MPOA effort is now
about 99% complete.

The following internet drafts were presented and discussed:

1. Transient Neighbors for IPv6 over ATM, Grenville Armitage et al,
   draft-armitage-ion-tn

This was the first of two proposals for IPv6 neighbor discovery over
ATM.  The goal was to have a single approach identified with which to
proceed.  This draft is a revised draft based on collaborative efforts
after the Montreal IETF, to combine two of the three drafts presented
there. The proposal calls for the MARS Cluster Control VC to be used
to multicast IPv6 Neighbor Discovery packets within a LIS, and NHRP is
used between LISes in conjunction with Neighbor Discovery redirects.
There were varied opinions expressed on the use of ND redirect,
several suggested improvements: the addition on the capability to
specify "this is the address I want to create a short cut to", and
clarifications on some of the diagrams.  Grenville is going to update
the draft to reflect his talk and the discussion.  This draft is now
the current working group consensus for IPv6 neighbor discovery over
ATM.

2. IPv6 over NBMA Networks, Ran Atkinson, draft-ietf-ion-ipv6-nbma

This draft was not presented.

3. RFC 1577 Update, Mark Laubach and Joel Halpern,
   draft-ietf-ion-classic2

Joel Halpern presented the current status of the draft.  It correct
all known problems with RFC 1577 behaviors, provides hooks for
migration to NHRP, and folds in the MTU material.  There will shortly
be a working group last call on the draft.

4. Classical IP to NHRP Transition Plan, Jim Luciani,
   draft-ietf-ion-transition

Jim discussed how Classical IP over ATM networks can be gracefully
transitioned to using NHRP. The key points are:

* The problem was simplified by not requiring ATMARP and NHRP servers
  to co-exist in the same LIS; instead, NHRP servers implement an
  ATMARP interface, and a single shared translation database will be
  used.

* ATMARP clients will not require any changes.

* ATMARP queries will generate responses derived from ATMARP only,
  while NHRP responses may be generated based on data learned via
  either ATMARP or NHRP.

Jim will update the draft.

5. NHRP Applicability Statement, Derya Cansever,
   draft-ietf-ion-nhrp-app

Derya presented the current version of the applicability statement.  A
questioner asked for an example of when NHRP should be used (answer:
when a particular link-layer QoS would be preferable to best-effort
router-to-router service), and there were several suggestions for
improving the document: adding a discussion of the looping problems
described in the router-router NHRP document, and adding a discussion
of NHRP's scaling problems for IP multicast.  Derya will update the
draft.  Masataka Ohta also volunteered to begin work on a separate
informational document on NHRP scalability.

6. Server Cache Synchronization Protocol (SCSP), Jim Luciani,
   draft-luciani-rolc-scsp

Comments from the mailing list and the previous ion meeting have been
added to the specification.  The portions of the document that
discussed MARS/MCS have been removed; MARS and MCSes will be discussed
in a separate document.  A last call will be issued on the draft.

7. Multicast Synchronization Protocol (MSP), Eric Mannie,
   draft-mannie-stc-ion-msp

Eric presented MSP, which is an alternative server synchronization
protocol.  During the Q&A period it was observed that this proposal
did not provide new functionality over SCSP or a compelling reason for
the WG to switch to its use.

8. Redundant MARS architectures and SCSP, Grenville Armitage,
   draft-armitage-ion-mars-scsp

Grenville presented an overview of the issues surrounding
synchronization of multiple MARS servers. A few significant points
were:

* Clients should not require modification and should "see" a single
  server at any given time.

* Fault tolerance. The current MARS specification provides an
  architecture for basic fault tolerance, although there may be race
  conditions.

* Load Sharing. In large clouds it would be useful to distribute the
  MARS server functionality.

9. VENUS - Very Extensive Non-Unicast Service, Grenville Armitage,
   draft-armitage-ion-venus

Grenville Armitage presented his VENUS document. He discussed "from
MARS to VENUS, and points in between." His major point is that
multicast shortcuts are a significant problem, and should not be used
until better understood.  A working group last call will be issued on
progressing this document as an informational RFC.

10. Multicast Server Architectures for MARS-based ATM Multicasting,
    Rajesh Talpade, draft-ietf-ion-marsmcs

Rajesh presented his draft; there was very little discussion following
his presentation. The intention is to issue a working group last call
on progressing this document as an information RFC.

11. EARTH - EAsy IP multicast Routing THrough ATM clouds, Michael
    Smirnov, draft-smirnov-ion-earth

Michael presented EARTH, which "is positioned between MARS and VENUS".
During the discussion, the point was made that Michael seemed to have
duplicated some of MARS' functionality, and the differences between
his draft and MARS were clarified.  He will update the document based
upon the discussion.

12. Another ATM Signaling Protocol for IP (IP-SVC), Kenji Fujikawa,
    draft-fujikawa-ipsvc

Kenji presented his draft, which outlines the results of his research
into an alternative ATM signaling protocol.  The protocol was designed
to be simple, useful in small LANs, and support auto-configuration.
Although this was thought to be an interesting idea, there were
concerns about the limited scalability of the approach.  In addition,
it is explicit in the ION working group charter that the group may not
work on ATM-layer signaling.  It was decided to continue discussion on
the future of this document on the mailing list, since it would
require WG consensus (and the agreement of the area directors) to
change the WG's charter to include this work.

End of minutes
________________________________________________________________________
Andrew G. Malis   malis@casc.com   phone:508 952-7414   fax:508 392-9250
Cascade Communications Corp.      5 Carlisle Road     Westford, MA 01886