Editor's Note: Minutes received 11/25/92

CURRENT_MEETING_REPORT_


Reported by Paul Tsuchiya/Bellcore

Minutes of the Inter-Domain Multicast Routing BOF (IDMR)

Agenda


  1. Determine the Charter of the Group.
  2. Go over the CBT (Core Based Trees), a proposal for scalable
     multicast routing.


The first IDMR Bof was held November 17th at the Washington, DC IETF
meeting.  It was chaired by Tony Ballardie of UCL. Tony amply
demonstrated early on in the meeting that the English had best stick to
dry humor, and leave the wacky stuff to the Americans (MP
notwithstanding).  Paul Tsuchiya was scribe, and these Minutes were
written by him.

Concerning Agenda item 1, it was agreed that there is a need for a new
multicast protocol for inter-domain multicast, as the existing schemes
do not scale well enough.  Therefore, it was decided that:


  1. A Working Group should be formed (the IDMR Working Group), and
  2. The charter of the Group is to design a standard multicast routing
     protocol for inter-domain multicast routing.


Though there was no explicit call for consensus, it was assumed that
Tony Ballardie would Chair the Group, with Paul Tsuchiya as alternate
Chair.  I assume that it is still possible for people to volunteer to
Chair the Group.  Also, there was no consensus (for or against) that CBT
should serve as the base text for the new IDMR protocol.  On the other
hand, no other proposals are on the table.

Concerning Agenda item 2, the following concerns were raised about CBT.


   o There was a concern that the current Internet-Draft didn't
     adequately specify the case where a node receives two joint
     requests before getting back the first ack.

   o The choice of which major core to send terminate request messages
     to is not specified correctly.

   o It was felt that there must be a way for detecting the case where
     there are two cores with uptree links on the same LAN.

   o The format for the core list packet must be worked out, and the

                                   1





     drawing in the CBT document is hard to understand.

   o There was a concern that the text describing when to send a quit
     request was not complete.

   o The notion of sending a quit some time after receiving a join ack
     (when changing parents) is no good (should send quit immediately).

   o There was a lot of discussion about what to do when the link to the
     parent goes down.  This whole issue needs to be worked out, but
     there seemed to be a general preference for flushing the whole tree
     below the break, with everything below subsequently rejoining.


Attendees
 
Anthony Ballardie        A.Ballardie@cs.ucl.ac.uk
Tony Bates               t.bates@nosc.ja.net
Scott Brim               Scott_Brim@cornell.edu
Michael Collins          collinsms@es.net
Barbara Denny            denny@erg.sri.com
Hans Eriksson            hans@sics.se
Roger Fajman             raf@cu.nih.gov
Dino Farinacci           dino@cisco.com
Shoji Fukutomi           fuku@furukawa.co.jp
Joel Halpern             jmh@network.com
Don Hoffman              don.hoffman@eng.sun.com
Dwight Jamieson          djamies@bnr.ca
Oliver Jones             oj@pictel.com
Paulina Knibbe           knibbe@cisco.com
Jim Knowles              jknowles@binky.arc.nasa.gov
John Krawczyk            jkrawczy@wellfleet.com
Padma Krishnaswamy       kri@sabre.bellcore.com
Greg Minshall            minshall@wc.novell.com
John Moy                 jmoy@proteon.com
Jim Perchik              perchik@athena.mit.edu
Thomas Pusateri          pusateri@cs.duke.edu
Benny Rodrig             4373580@mcimail.com
Henning Schulzrinne      hgs@research.att.com
Martha Steenstrup        msteenst@bbn.com
Tang Tang                tt@virginia.edu
Paul Tsuchiya            tsuchiya@thumper.bellcore.com
Ioannis Viniotis         candice@ececho.ncsu.edu