[MMUSIC] BUNDLE: Session bandwidth (Re: comments on draft-ietf-mmusic-sdp-bundle-negotiation-01)

[Harald Alvestrand <[e-mail]>]

On 11/04/2012 07:23 AM, Charles Eckel (eckelcu) wrote:
> 2) Section 6.4.2, reads:
>
>     The total proposed bandwidth is the sum of the proposed bandwidth for
>     each "m=" line associated with a negotiated BUNDLE group.
>
> I'm afraid this is an overly simplistic approach. For example, a typical video call with content sharing includes one audio m-line and 2 video m-lines. The bandwidth is declared at the session level as well as at the media level, but the sum of the media levels is typically greater than that of the session level. When content sharing occurs, it steals some bandwidth from the main video. In the case of BUNDLED m-lines, the proposed bandwidth should therefore be the lesser of the sum of the BUNDLE m-lines and the session bandwidth.
>
I'm not sure this really matters; what we're trying to compute is the b= 
parameter for the "implied RTP session" in the BUNDLE group. What 
matters is that it's computed consistently, which argues for the 
simplest possible computation.

Last time I looked b=CT was scoped to a whole conference ("supplied for 
the session"), while b=AS was scoped to a single RTP session. But I may 
be misinterpreting RFC 4566 section 5.8.

But yes, that's one of the advantages of the MMT approach - the sender 
gets to specify instead of the receiver having to compute.