Re: [AVTCORE] RTP Internal Multiplexing of RTP sessions

[Harald Alvestrand <[e-mail]>]

Speaking as RTCWEB contributor....

I think the discussion in RTCWEB has shown clearly that there isn't a 
clear consensus in the community on what an RTP session is, what the 
existence of RTP sessions requires from the applications, and what the 
RTP session is an useful tool for.

The debate has been framed in terms of the cost (in both complexity and 
network resources) of running multiple transports when no difference in 
treatment is desired from the network - but I think the confusion goes 
deeper than that; in the discussions in the W3C WEBRTC WG, which deals 
with the API requirements for the same thing, there doesn't seem to be 
any sign that anyone sees the need to represent an "RTP session" at the 
API offered to the Web applications using RTCWEB.

I'm looking forward to the discussion, both inside and outside the 
AVTCORE session, in Quebec.

                      Harald

On 07/19/11 17:35, Magnus Westerlund wrote:
> WG,
>
> As Individual contributor.
>
> I don't know if people in the WG are aware of the discussion ongoing in
> RTCWEB around multiplexing of protocols over the same underlying
> transport flow, like UDP flow. That include RTP sessions.
>
> There is a proposal:
> "Multiplexing of Real-Time Transport Protocol (RTP) Traffic for Browser
>                    based Real-Time Communications (RTC)"
> https://datatracker.ietf.org/doc/draft-rosenberg-rtcweb-rtpmux/?include_text=1
>
> This proposal is criticized by the following draft.
> https://datatracker.ietf.org/doc/draft-perkins-rtcweb-rtp-usage/?include_text=1
>
> There is an ongoing discussion on several aspects of these proposals on
> the RTCWEB WG. You are welcome to go join the discussion of the other
> questions raised on that mailing list.
>
> Here I primarily want to raise the issue I see with multiplexing several
> RTP sessions onto the same lower layer transport without something
> explicitly identifying these separate sessions. And in particular the
> RTP SSRC semantics changes that is proposed by the Rosenberg draft.
>
> Personally I think it is time we tackle the issue of how to efficiently
> run multiple RTP sessions over one lower layer transport. This as there
> are several wanting to have that functionality. However, I think this
> must be done in a very careful way. As we don't want to create an
> alternative version of RTP which is what the Rosenberg proposal in fact
> does. I do want us to gather the requirements on this and think of how
> we can accomplish this.
>
> I think there is a clear risk that we will not be able to accomplish it
> without either adding some type of shim header which has it sets of
> drawback or in fact fork or change the RTP protocol in such a way that
> it will not be fully compatible with what exist today. But if that is to
> be done, we should have a very clear consensus on that and ensure that
> we then develop future extensions to RTP within a single version of the
> protocol, not two.
>
> As WG chair:
>
> We are considering re-arranging the agenda to facilitate a debate of
> this topic in Quebec.
>
> So please indicate both your view of the topic in general and if you
> think we should re-prioritize the agenda.
>
> cheers
>
> Magnus Westerlund
>
> ----------------------------------------------------------------------
> Multimedia Technologies, Ericsson Research EAB/TVM
> ----------------------------------------------------------------------
> Ericsson AB                | Phone  [phone]
> Färögatan 6                | Mobile [phone]
> SE-164 80 Stockholm, Sweden| mailto: [e-mail]
> ----------------------------------------------------------------------
>
> _______________________________________________
> Audio/Video Transport Core Maintenance
> [e-mail]
> https://www.ietf.org/mailman/listinfo/avt
>