Re: [MMUSIC] BUNDLE: mandate RTP/RTCP multiplexing?

Hi,

Things would certainly work with separate RTCP ports. But, it would mean an additional 5-tuple, with associated ICE etc, which is what we want to avoid by using BUNDLE.

And, as BUNDLE already mandates the SDP offerer to offer rtcp-mux, my suggestion is to mandate the SDP answerer to accept it (assuming it enables BUNDLE, that is).

However, if people have issues with a MUST, a compromise would be “SHOULD, unless some mechanism explicitly requires the usage of separate RTCP ports”… or something like that.

Regards,

Christer

From: mmusic-bounces@ietf.org [mailto:mmusic-bounces@ietf.org] On Behalf Of Justin Uberti
Sent: 2. toukokuuta 2013 1:29
To: Paul Kyzivat
Cc: mmusic@ietf.org
Subject: Re: [MMUSIC] BUNDLE: mandate RTP/RTCP multiplexing?

+1 to mandating rtcp-mux when BUNDLEing.

Otherwise, we have weird situations, such as if you try to BUNDLE a SCTP session and a non-RTCP-mux RTP session together; the former has 1 ICE component, the latter has 2.

On Tue, Apr 30, 2013 at 12:30 PM, Paul Kyzivat <pkyzivat@alum.mit.edu<mailto:pkyzivat@alum.mit.edu>> wrote:
On 4/30/13 3:00 PM, Martin Thomson wrote:
On 30 April 2013 10:12, Paul Kyzivat <pkyzivat@alum.mit.edu<mailto:pkyzivat@alum.mit.edu>> wrote:
What are the implications of *not* mandating it?

I'm more interesting in learning why someone might want to split it
out.  I can't see any reason for doing that...

...that is, except perhaps for some of the fictitious attacks I invent
when someone suggests that RTCP is a good place for a security
mechanism of one sort or other.

Well, IIUC it was separate before there was a way to combine them.
Why was that? Why hasn't separate RTCP gone away entirely?

        Thanks,
        Paul

_______________________________________________
mmusic mailing list
mmusic@ietf.org<mailto:mmusic@ietf.org>
https://www.ietf.org/mailman/listinfo/mmusic