Small Group Multicast BOF (sgm)

Monday, July 31 at 1300-1500
============================

CHAIR: Randy Bush <randy@psg.com>

DESCRIPTION:

A number of people in the Internet community have been advocating that
the IETF consider standardizing a method for handling multicast to 
small groups, where "small" is generally understood to be in the range
of 3-10 or so participants. The rationale for this is twofold:

1) there are alleged to be a number of very important applications 
   which have a small numbers of participants per group (e.g. classic
   audio and video conferencing, collaboration tools, etc.) with an 
   extremely large number of such groups simultaneously active on the
   Internet.

2) Existing IP multicast methods, including newer specialized schemes
   like Source Specific Multicast (SSM) are generally thought to have 
   scaling problems when there are a very large nubmer of groups, 
   because of per-group state in all the intermediate routers.

Because of the range of possible approaches to this problem and the
potential widespread impact on portions of the IP protocol suite, 
router implementation, operational practices, etc. of the Internet, 
the BoF is being jointly hosted by the Routing and Internet areas, 
and will be chaired by Randy Bush of the OPS area.

The BoF's goals are to:
a) assess interest in forming a Working Group on Small Group Multicast
b) produce a crisp problem statement which could be used as the
   basis for a Working Group Charter
c) reach consensus on a set of constraints on the solution space (see
   below) d) examine existing work for possible solutions (see below)

In order to have a technically sound and deployable solution, it is
particularly important the scope of changes to the Internet 
architecturte, protocols, host & router implementations, and 
operational management be understood and hopefully minimized. Some of
these possible constraints include:

- Host API (e.g. sockets interface) changes allowed?
- Host protocol implementation changes allowed?
- IP packet header changes allowed? If so, effect on router forwarding
  speed?
- New routing protocols needed? Modifications to existing multicast
  and/or unicast routing protocols?

In order to stimulate discussion in advance of the BoF, we will use an
existing mailing list set up by the folks who have been working on a 
set of approaches to SGM which are collectively known as explicit 
multicast, or "xcast". The mailing list is:
        xcast@public.alcatel.com
and may be subscribed to by via xcast-reqeust@public.alcatel.com

A detailed agenda will be provided in the next couple of weeks after 
initial discussion on the mailing list.

An initial reading list for people interested in the discussion 
follows:

1) "Connectionless  Multicast",  draft-ooms-cl-multicast-01.txt
2) "Small Group Multicast",
   http://www.icair.org/main_projects_infrast_sgm.html
3) "Multiple Destination  option  on  IPv6",  draft-imai-mdo6-00.txt
4) "Distributed Core Multicast: a  routing  protocol  for
   many small groups with application  to  mobile  IP
   telephony", draft-blazevic-dcm-mobility-00.txt
5) "Datagram  Routing  for  Internet  Multicasting",  L.
   Aguilar, Sigcomm84, March 1984
6) "IPv4 option  for  Sender  Directed  Multi-Destination
   Delivery", C. Graff, RFC1770

AGENDA:

-  Intro / Why are we here     Randy Bush           5 minutes
-  Overview of SGM/Xcast       Dirk Ooms           10 minutes
-  CLM                         Dirk Ooms           15 minutes
-  SGM                         Rick Boivie         15 minutes
-  MDO6                        Yuji Imai           15 minutes
-  Xcast & SIP                 Bart Van Doorselaer  5 minutes
-  Security for Xcast          Olivier Paridaens   10 minutes
-  Args for & against SGM/Xcast   Rick Boivie      15 minutes
-  Discussion                  All                 30 minutes