Re: [rtcweb] draft-alvestrand-one-rtp-00.txt and intermediaires

[Christer Holmberg <[e-mail]>]

Hi,

>>When putting the draft together, did you consider intermediaires that do not support the mechanism? Even if the >>endpoints support it, intermediaries would still see and expect separate streams.
>
>I had not really considered this .... I am not convinced it is a problem.
>Brief analysis:
>
>Either they are in the media path or they are not in the media path.
>
>If they are not in the media path, they will not see the difference, and will not care.
>
>If they are in the media path.... we have a problem, but that same problem applies to any other SDP extension that >affects the media, such as RTCP multiplexing; if that's successful, that's a hopeful sign that they either don't exist or >aren't important.

I guess the question is what the impacts are.

In this case, the SDP offer/answer as such may go fine, but the intermediary might reserve bandwidth only for a single media type (the one which the SDP answer indicates will eventually be used for the multiplex) - even if that media type will consist of multiple multiplexed media types. 

Regards,

Christer