H323/SIP Interworking - way forward

[Christian Groves]

G'Day all,

The [H.323-SIP/Internet] thread has lead to some interesting discussion.
Being personally involved in the H248 "success" and having actually
having had to fight through the process of having to work between SG16
and IETF I think there's a few steps that need to be taken before we go
into mapping information elements.

I believe the first thing we need to do is define what architecture and
call scenarios we are working with. H323 has several architectural
elements, SIP is not so strong on architecture so we need to show what
elements are talking to each other. Only then we can define what
interfaces need to be interworked. In H248 this was relatively easy as
the interface was between an MGC and MG. The H323-SIP interworking would
be another degree of difficulty as there are potentially several
interfaces. This architecture work would not slow down the work but
would lead to a structured way of working were people agree on a problem
space. One of the issues with H248 was that the problem space was not
agree on and we went back and forth trying to figure out where we were.

As I mentioned at the SG16 I invite people to make contributions on
this. Anything more detailed at this stage I believe is counter
productive. Even Orit's first proposal I think is too detailed at this
stage. Look at the early H248 work where we had many detailed protocols.
There was an inordinate amount of work spent (and many late nights)
trying to get from details to architecture. Then we had to go back again
and define the protocol.

The 2nd thing to do is to define the set of capabilities we intend to
interwork. This needs to be a clear statement of scope. Interworking
basic audio with no conferencing and services is significantly different
than interworking full audio, video and data conferencing. In H248 the
scope was so large that people were working on completely different
things without appreciating the other people's requirements.

Thirdly, we must choose our base. What H323version will be interwork
with what SIP flavour and functionalities? Of course in today's world
there needs to be flexibility to address new functionality but we
shouldn't add a new work item every time of version 0 of an internet
draft from Joe Bloggs comes out.

Fourthly, I believe that once the above is done, then we have a very
good idea of what functional interworking work needs to be done. Once we
have an idea of the work we can also figure out where the
responsibilities would lie.

Once we know the work and who is responsible for the work then we MUST
make sure this is agreed upon and documented in both the ITU and IETF by
the people who can make decisions. Too many things were hidden in the
H248 work and people that could make decisions were not at the
appropriate meetings.

Finally we can then produce the detailed documents in a way that
everyone understands.

I do think that at the end of the day H.248 was a success and that the
co-operation between the IETF and ITU was made to work. This was because
of the large amount of work and sacrifices made by a few individuals and
companies who wanted to make this happen. Many of the individuals
working on H248 will probably contribute on H323/SIP interworking so
lets learn from the mistakes and have another go at making H323-SIP
interworking work without having to kill ourselves to achieve the goal.

Cheers, Christian