Re: [rtcweb] New use case: Large multiparty session

[Harald Alvestrand <[e-mail]>]

On 08/03/11 15:34, Arno Bakker wrote:
> On 03/08/2011 14:51, Randell Jesup wrote:
>> On 8/3/2011 4:36 AM, Arno Bakker wrote:
> >>
>>> assume a client is involved in a large multiparty session but is
>>> unable (or unwilling) to upload his signals to all participants.
>>>
>>> This suggests forwarding support from the network is required, e.g.
>>> some relay server, multicast, or the future IETF peer-to-peer streaming
>>> protocol (PPSP).
>>
>> So the use case is:
>>
>> User is part of a multiparty (3 or greater) session, but for one of
>> several reasons (such as
>> available upstream bandwidth) cannot send media to all other
>> participants ("mesh" conferencing).
>> This requires that another node, either a central network node (such as
>> a conferencing server)
>> or another member in the session, relay or mix the user's media to
>> distribute to the rest of the
>> members of the session.
>>
>
>
> Yes. Although your wording suggests a single relay node (central or 
> member) is used. Please leave the option of true peer-to-peer 
> distribution (or IP multicast) of the signal open.
In draft-ietf-rtcweb-use-cases-and-requirements-01.txt, the "central 
server" model is listed in section 4.3.3, "Video conferencing system 
with central server".

I am strongly opposed to adding use cases (which in turn generate 
requirements) to this first iteration of RTCWEB that we do not know how 
to fulfil with any currently deployed standard technology, so I would 
definitely want to leave the "peer to peer distribution or IP multicast" 
out of the picture for now. It can be added to a working appendix of 
"Use cases that have been discussed, but will not be considered as 
requirements in this iteration of RTCWEB".