Ad Hoc: Conferencing Issues Date: 20 March 2002 Location: Hilton Harmony
Moderator: Eric Burger <eburger@snowshore.com> Scribes: Tim Melanchuk <timm@convedia.com>, Roni Even <roni.even@polycom.co.il>
Summary: We will start the discussion of conferencing requirements in the SIPPING group. Work may end up in SIPPING, SIP, MMUSIC, or a new work group. We will develop a requirements document that will include requirements, the model and the terminology. The document will include all the requirements, such as call control as well as floor control. The end-goal is to have a BCP RFC that will point to other relevant drafts.
The model addressed in the document will be centralized conference control (star configuration), with both centralized or end point mixed media. We decided that full mesh control is nice but not practical.
A first draft of the document should be ready in three weeks. It will be written by Roni Even, Orit Levin, Daniel Petrie, and Andrew Zmolek.
Minutes of the conferencing adhoc: March 20, 2002
- motivation: there are 5 or 6 active drafts which discuss aspects of conferencing and which span multiple work groups, most notably drafts presented in the sipping meeting by Orit Levin, Rohan Mahy and Henning Schulzrinne.
- question: should this work be carried be carried out in MMUSIC which has charter elements which address conferencing; SIP, where much of the current application of conferencing work is taking place; SIPPING, which is chartered to define new SIP requirements; or a new WG because none of the above is a perfect fit?
- answer: keep in SIPPING until specific requirements can be articulated which can be moved off to either more appropriate or a new WGs. This may require one or more BCP RFCs to tie together.
- question: what is conferencing?
- discussion: there are different definitions and models captured in the various drafts. These need to be coalesced into a unified model with consistent terminology. There was a general consensus that this should be possible although we can't solve all of the problems at the same time. It is difficult to go down the path of arbitrary complex graphs.
- direction from the ADs is that at least two topologies must be supported; full mesh with a small number of participants (~3-5 with endpoint mixing) and star
- question: can conferences with the two topologies be joined? answer: no.
- question: can a conference transition between the 2 topologies? answer: yes.
- question: what is meant when we talk about topologies?
- discussion raised the following (possibly contradictory points):
- we are only talking about control; media is defined by the SDP - mesh based mixing won't work on a large scale - different control topologies are possible but every endpoint (except an MCU) sends unmixed media - a 3-way conference can be a star; every endpoint can do its own mixing in an IP network (cascade model) - if you start with a star, it be moved/grown - in a star, can there be a single master for control where it is transferred if the master drops out - floor control topology may be different from that of call control - floor control may be similar to IM multi-party conversations (state is shared by some magic means to all participants) - transport of floor control messages is separate from call control
conclusion: there are different graphs/topologies for different resources (call control, floor control, media, others?). Full mesh control is too hard and there should be some centralization of control. However different topologies, from mesh to centralized, can be used for media.
- A mesh of stars, or cascading conferencing domains, is explicitly out of scope.
- going forward: the following four people have volunteered to create a single requirements/framework/architecture document in the next 3 weeks:
- Andrew Zmolek: zmolek@avaya.com - Roni Even: roni.even@polycom.co.il - Orit Levin: orit@radvision.com - Daniel Petrie: dpetrie@pingtel.com
Conferencing Ad Hoc Participants:
Eric Burger eburger@snowshore.com Orit Levin orit@radvision.com Daniel Petrie dpetrie@pingtel.com Paul Rupsis paul.rupsis@intel.com Peter Blatherwick peter_blatherwick@mitel.com Alan Johnston alan_johnston@wcom.com Francois Audet audet@nortelnetworks.com Tim Melanchuk timm@convedia.com Roni Even roni.even@polycom.co.il Brian Rosen brian.rosen@marconi.com Sean Olson seanoolson@yahoo.com Magnus Westerlund magnus.westerlund@ericsson.com Don Stanwyck don@stanwyck.com Andy Zmolek zmolek@avaya.com Mary Barnes mbarnes@nortelnetworks.com Sriram Peremswan sriramp@nortelnetworks.com Cullen Jennings fluffy@cisco.com Joerg Ott jo@ipdialog.com Colin Perkins csp@isi.edu Peleka Pessi peleka.pessi@nokia.com Eunsook Kim eunak@etri.re.kr
updated 21 March 2002 21:05 -0600