Re: [rtcweb] How to multiplex between peers

On Wed, Jul 20, 2011 at 10:18 AM, Bernard Aboba
<bernard_aboba@hotmail.com>wrote:

>
> > Timing is a real problem. If we want to do multiplexing in RTCWEB 1.0 we
> > need to pick what is ready and available. We can't go around proposing
> > new things that are complicated. Changing the SSRC field semantics in
> > RTP/RTCP is a complicated thing.
>
> [BA] We need to ask ourselves "Does this really need to be in the critical
> path for RTCWEB 1.0"?
> As I understand it, this would be a negotiated feature, and presumably all
> implementations
> would be required to support RTP/RTCP as defined in RFC 3550 and suitable
> profiles.
> So from my perspective, I'd prefer to see this handled as an extension on
> its own path, not a
> mandatory-to-implement feature in 1.0.
>

This is where I've ended up on this topic. We can easily multiplex multiple
RTP sources (of the same type) over a single RTP session using SSRC. We can
also mux RTCP over the RTP session using RTCP mux. So, for an arbitrary
video call, we have just 2 RTP sessions/NAT bindings.

Is it worth going the extra mile to get down to 1 in v1.0, given the lack of
consensus that exists right now? Is there even a compelling argument to do
so?

>
>
> > Isn't that gateway going to be present anyway in most case. I think the
> > ICE connectivity checks are one thing that will force a gateway in many
> > cases. In the cases it isn't needed, then you can skip the multiplexing
> > in that case to reach the legacy.
>
> [BA] Just because we're likely to require a gateway doesn't mean we should
> heap more and more things on the camel's back.  We're already talking about
> ICE, shims, transcoding, changes to RFC 3550....  that's a lot of
> functionality
> for the "gateway".  So I agree with Jonathan about the disadvantages of a
> shim
> (or DCCP for that matter).
>
>
> > The big difference is that we either have a well reviewed and by the
> > community and IESG approved specification in DCCP. Or we create
> > something our selves that needs to go through all that review and will
> > be even less tested than DCCP is.
>
> [BA] Deployment counts, too.  How widely deployed is DCCP for realtime
> uses?
>
>
>
> _______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb
>
>