home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Internet Info 1997 December
/
Internet_Info_CD-ROM_Walnut_Creek_December_1997.iso
/
ietf
/
mmusic
/
mmusic-minutes-95jul.txt
< prev
next >
Wrap
Text File
|
1995-10-18
|
5KB
|
141 lines
CURRENT_MEETING_REPORT_
Reported by Mark Handley/University College London
Minutes of the Multiparty Multimedia Session Control Working Group
(MMUSIC)
An on-line copy of the minutes and the accompanying PostScript slides
are available from ftp://ftp.isi.edu/confctrl/minutes in the files
ietf.7.95 and slides.7.95.tar.
The Multiparty Multimedia Session Control Working Group (MMUSIC) met for
two sessions on July 17 and 18 at the 33rd IETF.
Interworking With The ITU
Carsten Bormann gave a presentation (slides.7.95.a) of the possibilities
for interworking between the ITU T.120 family for multimedia
conferencing and Internet-based approaches to conferencing. The ITU
appears to be moving towards using RTP or an RTP-like protocol for media
data streams, but it is unclear how the conference control functionality
encompassed in the T.120 suite would be achieved over the wide area
Internet. Initially the ITU seems to be interested only in LAN
interoperability.
Three interoperability scenarios were presented, and the implications
examined:
1. T.120 terminal phones in to ``classical'' IETF multicast session
2. Internet protocols only used locally behind a gateway
3. Tight interworking with the illusion of homogeneity
The issues where there are potential incompatibilities include:
o Resource control
o Addressing: use of multicast
o Session setup: prescriptive session description vs. negotiated
o Security: cryptography vs. link setup authentication
The details of these interoperability scenarios are discussed in more
detail in draft-bormann-mmusic-itu-interop-00.txt.
This presentation was followed by a discussion of how the MMUSIC Working
Group should relate to the T.120 work. The rough consensus was that the
group should pay attention to the ITU work, and not needlessly replicate
T.120's conference control functionality in an incompatible way.
However, the opinion was expressed that the group's current emphasis on
wide area scalable multicast-based conferencing was desirable, and that
we shouldn't sacrifice the benefits of multicast based sessions to
conform to the ITU tightly coupled model. No real conclusion was
reached, except that a building-blocks approach was advocated.
Vertical Control Protocols
Joerg Ott gave a presentation (slides.7.95.b) of two vertical control
protocols (control protocols between local applications and controllers)
developed at TU Berlin: The LASSO Dictionary and the Conference
Information and Request Protocol (CIRP). Unlike CCCP and the control
architecture described by Henning Schulzrinne (see minutes.4.95), these
consist of a central (for each participant) information store, and an
access protocol to this data store. The dictionaries object store is an
acyclic directed graph, which is sufficient to reflect T.120 conference
data, as well as more loosely couple conferences. CIRP complements the
dictionary and allows a more flexible asynchronous event triggered
interaction style, but still requires a single conference controller per
user. Work is currently in progress to unify the dictionary and CIRP.
Changes Made To The Session Description Protocol
Mark Handley presented the changes that have been made to the Session
Description Protocol since the Danvers IETF (slides.7.95.c). The
current version of the draft is available from
http://www.cs.ucl.ac.uk/staff/M.Handley/sdp.01.ps.
The main changes are:
o SDP is no longer a superset of the sd protocol.
o The ``c'' fields have been split into channel and time fields.
o Repeat time fields have been completely re-worked.
o Origin fields have been completely re-worked, and now uniquely
identify a session announcement.
o Bandwidth fields have been reworked somewhat.
o Media fields have been reworked somewhat.
o The Session Description Protocol has been separated into a separate
draft from the announcement protocol that carries it.
In general, the changes attempt to make the protocol more generic and
more widely applicable.
A number of minor problems with the protocol or the description in the
current draft were noted. All these changes are fairly minor, but the
most significant are:
o The ``o'' field allows spaces in the address. The ``c'' field does
not. The incompatibility should be resolved.
o For unicast, the existing ``c'' field may not be sufficient to
identify whether the address is to send to or receive from or some
other semantics. This should be investigated further.
o <address type> in ``o'' and ``c'' fields should probably be two
fields: <network type> and <address type>.
o For unicast, hierarchical encodings need to be able to use multiple
ports (for multicast they use multiple multicast addresses). The
notation <base address>/<number of addresses> should be extended
to ports in the media field.
A new draft will be released shortly addressing these issues. Assuming
this does not raise new objections, the consensus of the group was that
we should submit SDP for consideration as a Proposed Standard around the
time of the Dallas IETF.
Other Business
It was flagged that the Agreement Protocol is now available as an
Internet-Draft: draft-ietf-mmusic-agree.txt,ps