Re: [rtcweb] SAVPF history (Re: Final plea about SRTP)

[Magnus Westerlund <[e-mail]>]

On 2012-05-08 15:34, Roman Shpount wrote:

> I would say that something that changes to rfc4585 would be more
> relevant. Regardless of the actual changes in the draft, my point is
> there are very few actual SIP devices that implement SAVPF or AVPF.
> Specifying this as the only profile supported by WebRTC will create yet
> another challenge to legacy interop, since it will require not only
> processing of ICE messages, and DTLS-SRTP re-encoding, but also RTCP
> re-generation.

Hi,

Here I have to protest that it increases the barrier. First of all the S
in SAVPF is for using SRTP. That is already agreed on in this WG.
Secondly the F stands for the feedback part. That consists of two parts:

1) New Timer Rules

2) Feedback RTCP packets

The second is actually negotiated on a per individual feedback format
basis. So only the first is what really what would be mandated by a
WebRTC requirement of SAVPF. From my perspective to get the RTP/RTCP
functionalities needed to give a good user experience we really need the
timing rules. When it comes to legacy interop a signalling gateway can
actually produce a SAVPF configuration that is fully interopable with an
SAVP end-point.

The configuration required to make this interoperable are: use a=rtcp-fb
to sett trr-int to 3.5 to 4 seconds and ensure that the same or at least
sufficient RTCP bandwidth is configured and that no feedback format are
agreed on being used.

So please don't be afraid of SAVPF.

Cheers

Magnus Westerlund

----------------------------------------------------------------------
Multimedia Technologies, Ericsson Research EAB/TVM
----------------------------------------------------------------------
Ericsson AB                | Phone  [phone]
Färögatan 6                | Mobile [phone]
SE-164 80 Stockholm, Sweden| mailto: [e-mail]
----------------------------------------------------------------------