Re: [MMUSIC] SCTP question: Where does it multiplex?

[Paul Kyzivat <[e-mail]>]

On 12/6/12 1:01 PM, Christer Holmberg wrote:
> Hi,
>
>> One way or another the SDP must define the SCTP association, and if it
>> is to share a 5-tuple then the grouping to do that needs to be defined.
>> IIUC, the current bundle draft isn't capable of doing that.
>
> The current bundle draft needs additional work, mainly regarding the handling of SDP attributes etc, even to share a 5-tuple for RTP.

The current draft requires the <proto> fields of all the bundled m-lines 
to be the same. It also strongly implies that all of them must be 
RTP-related. To include the SCTP m-line for support of the data 
channels, it will be necessary dissimilar <proto> fields to be bundled. 
But not all combinations are valid.

So it needs some additional text to explain which other cases may be 
bundled. The simple way out would be to simply enumerate the different 
combinations of <proto>s that may be bundled and any other constraints 
that may apply. Better would be to define some more general criteria for 
those that may be bundled, so that it might survive future extensions.

An obvious constraint is that there must be some way of demuxing them.

I am still trying to understand the details of how the DTLS/SRTP is 
multiplexed with RTP (actually I guess it will be UDP/TLS/RTP/SAVP(F)). 
Is there some place that explains that? RFC 5764 explains how RTP, DTLS, 
and STUN are demuxed for DTLS/SRTP. And 
draft-tuexen-tsvwg-sctp-dtls-encaps explains how to run SCTP over DTLS. 
But those two mappings don't seem to be compatible.

Has anyone written down a proposal for how that will work?

	Thanks,
	Paul