CURRENT_MEETING_REPORT_

Reported by Tony Ballardie/University College London

Minutes of the Inter-Domain Multicast Routing Working Group (IDMR)

The IDMR Working Group met twice at the 31st IETF in San Jose.  The
first session was held on Monday, 5 December and the second on Tuesday,
6 December.


Session 1

Van Jacobson described a transition architecture for the MBone in
general.  The first phase in this transition involves converting the
MBone to multi-level multicast.  Steve Deering is implementing a scheme,
due to Ajit Thyagarajan, whereby multicast packets are encapsulated with
domain addresses at the boundaries of domains, and whereby DVMRP routing
information exchanged across domain boundaries is in terms of these
higher level domain addresses.  In this phase, the level 1 routers
within domains could run either DVMRP, MOSPF, PIM, or any multicast
routing protocol, and the border routers would all communicate with each
other via the level 2 protocol; for phase 1 the level 2 protocol would
be DVMRP (carrying domain level information).  Eventually the PIM folks
propose to transition the level 2 routers to speak PIM because of its
preferred scaling properties over DVMRP's ``broadcast and prune''
mechanism.  However -- and this is the point of the transition
architecture -- the Level 1-to-Level 2 protocol interaction will be
defined such that level 1 routers only need to be upgraded once, and
need not be upgraded later to interoperate with the level 2 PIM
deployment.  Currently the detail of the Level 1-to-Level 2 protocol is
being worked out, and the multilevel DVMRP is being implemented.


Session 2

Tony Ballardie presented the CBT protocol as it currently stands.  This
was the first presentation of CBT to the group for nearly a year, when
there were several issues (such as LAN designated router election) that
needed re-working.

There was also a presentation by George Oliva and Chuck Graff from the
U.S. Army on how the Army intends to make use of Internet multicast
(utilizing mobile and satellite communication channels) in the future,
rather than proprietary protocols.  This talk was intended as an
introduction, showing the major role the Army can play in the
utilization of multicast communication, and indicated their intent to
actively participate in the IDMR working group.

Finally, there was some important discussion on how the group should
proceed with there being two protocols (PIM and CBT) under review by the
group.  Many felt that PIM is a superset of CBT, and therefore PIM
should be advanced on the standards track (it is desirable for several
reasons that only one protocol be advanced along the standards track).
However, there was also strong opinion that, because of the lack of
implementation experience, the decision should not yet be made.  It is
hoped that CBT implementation results will be available by the next
meeting, at which time the situation will be reviewed again.

The Routing Area Director suggested that the group should discuss,
between now and the next meeting, whether CBT and PIM are sufficiently
different, and whether each has significant advantages in its own right
over the other, to warrant the further advancement of both protocols by
the group.