CURRENT_MEETING_REPORT_
Reported by Howard Berkowitz/PSC International and
Andrew Malis/Ascom Nexion
Minutes of the Routing Over Large Clouds Working Group (ROLC)
The ROLC working group met in two sessions. The first session is
reported on in these minutes. The second session, which has a separate
set of minutes (following these), was a joint meeting of the ROLC and
IPATM Working Groups.
There were over 120 attendees at the first session.
Agenda
o Agenda bashing
No changes were made to the agenda.
o ATM Forum MPOA Update
Presented by George Swallow, cisco Systems, Chair, ATM Forum
Multiprotocol Over ATM Working Group
o NHRP implementation experience
o NHRP open issues and draft-ietf-rolc-nhrp-IV.txt
Presented by Dave Katz, cisco Systems
o NHRP Applicability Statement (draft-ietf-rolc-nhrp-appl-01.txt)
Presented by Derya Cansever, GTE Laboratories
o NHRP MIB (draft-ietf-rolc-nhrp-mib-00.txt)
Presented by Mike Patrick, Motorola ISG
o Status and Workplan
ATM Forum Multiprotocol over ATM (MPOA) Update
The initial BOF was held last September, and the Forum began a working
group in November before the last IETF. The scope and requirements are
complete.
The plan is to support MPOA with an Overlay Model comparable to ROLC. It
is intended to be compatible with integrated PNNI, and a future Peer
Model is not precluded. Any solution must interoperate with LAN
emulation and non-ATM devices.
In the reference model, the MPOA service cloud interconnects with LAN
emulation as a separate service over ATM. The exact nature of this
interconnection has yet to be defined.
It is intended to be ``a good citizen of the Internet.''
MPOA and NHRP relationship: An MPOA requirement is to build on NHRP.
There is a strong feeling this should be more general with regard to the
network layer, however, the group is concentrating on ATM, rather than
supporting general link layers.
NHRP Implementation Experience
No implementation experience was shared in the meeting. The chair said
there are at least two implementations in progress. cisco Systems
announced that they are one of the two, and plan to begin testing soon.
NHRP Open Issues and Draft
The new draft was distributed on the list as
draft-ietf-rolc-nhrp-IV.txt. An official -04 version will be submitted
as an Internet-Draft several weeks following the meeting, and including
all comments and changes from the meeting.
Dave Katz gave an overview of the changes to the specification. This is
the same list of changes he included in the introduction to draft -IV
itself.
Dave then led the discussion.
o The need was discussed for a separate document discussing the
router-to-router case, because of its complexity. Dave will start
a new Internet-Draft on the topic.
o NHRP servers along the path are now not allowed to cache NHRP
entries information unless the new bit is set indicating that the
information is stable.
o If metrics are not preserved between routers (e.g., in OSPF-BGP
interactions), then there is the potential for looping. Full
semantic preservation of metrics prevents the formation of such
loops. NHRP is also loop free when used in one AS, or the new
stable bit is set in the NHRP reply, or if stub networks have no
back doors.
o The goal that NHRP is usable in connectionless (SMDS) as well as
connection-oriented (ATM) environments.
o The chair encouraged people to comment on the looping problem (or
anything else, for that matter). Revision -04 may be subject to
the Last Call for Proposed Standard; comments are solicited
following its publication.
o Curtis Villamizar wants the specification to state things more
clearly. It is known that there are scenarios where routing loops
can exist, and they cannot all be solved. This should be mentioned
in the applicability statement.
o The question was asked if we are denigrating NHRP because some of
the actions routers take? The purpose of this working group is
large clouds; the solution is not viable if it only works over
limited topology. Dave Piscitello replied that if the limited
topology is widely used, then the solution is still useful. NHRP
will also be applicable to cases where there is no exterior
protocol.
NHRP Applicability Statement
Derya discussed the changes he has planned for the applicability
statement, draft-ietf-rolc-nhrp-appl-01.txt:
o Clarify the router-to-router case
o Not a replacement for routing protocols
o Clarification of potential looping cases
o Needs to be harmonized with NHRP-IV draft
The chair reminded the working group that the applicability statement
and protocol analysis must be submitted in order to standardize , as
well as the MIB and implementation experience when available.
NHRP MIB
Mike led a discussion of some issues with the current draft of the MIB,
draft-ietf-rolc-nhrp-mib-00.txt.
o The MIB has not yet been compiled.
o The section on cut-through circuits will be moved to the
applicability statement.
o Traps: customers like them, but engineers do not. The current
plan is to not use traps.
o Unnumbered links over ATM: Consensus that these need to be
supported. Every unnumbered link should have an ifIndex entry.
Consider OSPF methodology here.
o RFC 1573 logical interface addresses---is this being done? Mike
will query.
o The MIB does not allow a LIS to be implemented on multiple NHRP
network IDs. This was agreed to be acceptable assumption.
o May need additional indexing for QoS, and different MAC addresses
for different QoS. MAC address plus QoS forms a tuple (cf. SNA
Virtual Route).
Status and Workplan
The chair asked to be told, in public or private, of implementations
underway. Kanan Shah will write the forthcoming protocol analysis
document.
July 95: Discuss implementation experience; submit NHRP document
to the IESG as a Proposed Standard; continue to review companion
documents.
December 95: Submit companion documents to IESG.
The charter will be updated to reflect this new workplan.