Here are the minutes for the call today - if I have gotten anything wrong,
or missed anything, please let me know.
Nancy
Minutes of the Megaco/H.gcp call held May 4/99, 10-12 ET
----------------------------------------------------------------------------
-----------
Attendees: Participants from both the IETF Megaco WG, and ITU SG16 were
present on the call. A total of 27 people called in.
Results are summarized here, and are open for discussion on the mailing
lists. (Actions are prefaced by **.)
Results:
-----------
1: Multimedia/H.320 discussion
- John Seger presented his paper on multimedia contexts
- More detailed call flows are needed for the multimedia context and the
monomedia context cases.
- criteria:
- protocol needs to be optimized for the most common case
- for multimedia contexts, need to distinguish between media streams for
applying events to the right stream, and mode is per media stream
- for monomedia contexts, need to specify the media of the context, and
which terminations are linked across contexts, there are more objects to
handle, but they are simpler
** Tom Taylor will provide the signalling for one or more call flows
** John Seger and Brian Rosen (and anyone else who wants to!) will work
through the multimedia context and the monomedia context call flows using
the signalling Tom provides.
2: Context Parameter for "mixing" the media
- there was agreement that, independent of whether a context is mono or
multimedia, we need a context parameter indicating how media is to be mixed
within a context. For audio it would default to simple mixing, but other
choices might be offered, for video it could be based on current speaker,
etc. In any case, there needs to be a context parameter to hold this
information.
3: H.245 vs SDP
- Agree that both H.245 and SDP can provide the same degree of
expressiveness.
- Neither is complete as it is
- we are not using SDP headers, just SDP syntax + additions
- H.245 already has a way to express capabilities such as MG resource
constraints described as " I can support 4 G.723 codes, or 7 G.726 codes, or
some particular mix of the two" - we can take this and use it to write the
same thing in SDP syntax.
** Gur Kimchi will write up an email on capability exchanges and send it to
the Megaco mailing list. With his method, he does not think it is hard for
the MGC to track resource usage in the MG.
- discussed whether the MGC would need to look inside the SDP/H.245 syntax
before shipping it off to the next MGC or MG. Most said, yes, the MGC would
need to look into it - need to verify it, perhaps need to modify it. Then
the MGC tells the MG what to do. Do we need to allow some flexibility here
for MGs that want to decide some of this information themselves?
- Issue then becomes:
- is SDP or is H.245 better for making the MGC scaleable and fault
tolerant?
- What is the load on an MGC depending on the protocol choice? Need
concrete comparisons of H.245 syntax and SDP syntax.
1) what is the info the MG must tell the MGC
2) how to encode the info the MG must tell the MGC (i.e. Audit)
3) how to encode the info an MGC tells an MG (i.e. Add, Modify)
** How will we advance? There is interest in seeing what Christian Huitema
is working on in this area. Will anyone else prepare syntax comparisons?
- H.245 in the MG
There is a contribution into SG16 for an H.320 MG to contain H.245. This
contribution (as well as other SG16 contributions) will be made available to
the Megaco mailing list. Issues with this: How does the MGC remain aware of
the resources the MG is committing? Is the H.245 backhauled to the MGC or
does the MG make decisions and notify the MGC of its decisions?
4: Scripting
- briefly discussed: if an MG wants to run using Java scripts and
notifications, they can be written up in a Java scripting package. No need
to add anything new to the protocol. MGC finds out if an MG has that
scripting package using an Audit.
5: Logical names for terminations
- related to mono vs multi media contexts discussion - if a context is
multimedia, a physical termination need only live in one context, if a
context is monomedia, a physical termination may need to be virtually
present in more than one context.
- structure of the termination name can be duplicated using parameters
6: Handling stereo in monomedia contexts
- if all terminations involved are stereo, only need one context. If some
are mono and some are stereo, then need one context for left, and one for
right.
7: Sharing information from SG16 to the IETF
** Tom Taylor and Glen Freundlich will work out the best way for IETF Megaco
participants to have access to ITU-T SG16 contributions to the Santiago
meeting.
8: Next Audio Call
May 11/99 10am-12pm ET - call details will be posted to the Megaco and
ITU-SG16 mailing lists.
----------
--------------------------------------------------------------------------
Nancy M. Greene
Internet & Service Provider Networks, Nortel Networks
T:514-271-7221 (internal:ESN853-1077) E:ngreene@nortelnetworks.com
