Re: [MMUSIC] Adam Roach's Discuss on draft-ietf-mmusic-sdp-simulcast-12: (with DISCUSS and COMMENT)

[Magnus Westerlund <[e-mail]>]

Hi Adam,

I think we have one real issue  beyond the inconsistencies that clearly 
was not updated to the latest syntax and some other mistakes. Thanks for 
catching these.

Den 2018-06-20 kl. 04:43, skrev Adam Roach:
> Adam Roach has entered the following ballot position for
> draft-ietf-mmusic-sdp-simulcast-12: Discuss
>
> When responding, please keep the subject line intact and reply to all
> email addresses included in the To and CC lines. (Feel free to cut this
> introductory paragraph, however.)
>
>
> Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html
> for more information about IESG DISCUSS and COMMENT positions.
>
>
> The document, along with other ballot positions, can be found here:
> https://datatracker.ietf.org/doc/draft-ietf-mmusic-sdp-simulcast/
>
>
>
> ----------------------------------------------------------------------
> DISCUSS:
> ----------------------------------------------------------------------
>
> Thanks to everyone who put in the hard work to make this document happen. I have
> found a blocking issue that I believe should be easy to address; but (due to the
> potential impact on interoperability) document publication does need to be
> blocked pending resolution of this issue. The problem is that the SDP examples
> in this document are not consistent with the syntax and semantics defined in
> draft-ietf-mmusic and draft-ietf-avtext-rid, as described below.
>

Snipped a number of editorial things that I fully agree needs to be 
changed.

>
> Also, if LPC is intended to be used with the first RID (as is suggested by the
> text above the example), then your "pt" value needs to be "pt=99,102,98" --
> otherwise, the RID will prevent the use of PT 98. Remember: RID defines
> *restrictions*. If you say "pt" and then don't list a PT in it, then that
> missing PT is strictly forbidden from appearing with that RID.

Yes, but 98 will never occur in an RTP payload header in this 
configuration, it will only be encapsulated in the RED payload format 
(100). See the mmusic-RID drafts section 8.3 that contains this example 
focused on this particular issue.

>
> There is a similar problem with the video section, which needs to read as
> follows to match the explanatory text:
>
>     a=rid:1 send pt=103,105,107;max-width=1280;max-height=720;max-fps=30
>     a=rid:2 send pt=104,106,107;max-width=1280;max-height=720;max-fps=30
>     a=rid:3 send pt=103,105,107;max-width=640;max-height=360;max-br=300000
>     a=rid:4 send pt=104,106,107;max-width=640;max-height=360;max-br=300000

Here we get into the real issue as I see it. My interpretation of RID 
and what Simulcast says is that the RTX and flexfec streams related to 
any Simulcast streams are not indicated in the PT list for the RID. 
This, due to that they are separate RTP streams. And for example a 
Flex-Fec stream could be related to multiple simulcast streams' packets.

draft-ietf-mmusic-rid-15 Section 4 says:

    Implementations that use the "a=rid" parameter in SDP and that make
    use of redundancy RTP streams [RFC7656], e.g.  RTP RTX [RFC4588] or
    FEC [RFC5109], for any of the source RTP streams that have "a=rid"
    lines remaining after applying the rules in Section 6 and its
    subsections, MUST support the RepairedRtpStreamId SDES item described
    in [I-D.ietf-avtext-rid] for those redundancy RTP streams.
    RepairedRtpStreamId MUST be used for redundancy RTP streams to which
    it can be applied.

    The redundancy RTP
    stream MAY (but need not) have an "a=rid" line of its own, in which
    case the RtpStreamId SDES item value will be different from the
    corresponding source RTP stream.

    Section 8.3 which contains the rest of discussion of redundancy 
which is focused only on RED type
   formats.

    draft-ietf-mmusic-simulcast-12 does not contain any discussion 
beyond the example in 5.6.3 of redundancy.

So this only requires the use of RepairedRtpStreamId which the example 
does not include as an RTP header extension, I will fix this.

So maybe it should be a requirement to indicate in the RID which 
redundancy mechanisms that will be applied to the different simulcast 
streams, however currently this definition are in my view done only on 
media description (m=) level and thus in Figure 8 flexfec can be applied 
to any simulcast stream, while RTX is defined for retransmitting packets 
with payload type 103 and 104. And likely if this is to be address this 
is a change to mmusic-RID.



>
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
>

The editorials I have removed and will make.

>
> Figure 7:
>
>>   m=video 49602 RTP/AVPF 96 104
>>   a=mid:zen
>>   a=rtpmap:96 VP8/90000
>>   a=fmtp:96 max-fs=3600; max-fr=30
>>   a=rtpmap:104 rtx/90000
>>   a=fmtp:104 apt=96;rtx-time=200
>>   a=rid:1 send pt=96;max-fs=921600;max-fps=30
>>   a=rid:2 send pt=96;max-fs=614400;max-fps=15
>>   a=rid:3 send pt=96;max-fs=230400;max-fps=30
>>   a=extmap:1 urn:ietf:params:rtp-hdrext:sdes:mid
>>   a=extmap:2 urn:ietf:params:rtp-hdrext:sdes:RtpStreamId
>>   a=rtcp-fb:* ccm pause nowait
>>   a=simulcast:send 1;~2;~3
> While this isn't technically invalid (modulo the extmap lines), it's saying
> something a bit odd. Namely, it's offering to use RTX if the answerer doesn't
> understand RID, but forbidding the use of RTX if they do. You probably meant
> something more like:
>
>     a=rid:1 send pt=96,104;max-fs=921600;max-fps=30
>     a=rid:2 send pt=96,104;max-fs=614400;max-fps=15
>     a=rid:3 send pt=96,104;max-fs=230400;max-fps=30
>
> Or, really, since that's all the PTs that are defined for this media section,
> just leave it out altogether:
>
>     a=rid:1 send max-fs=921600;max-fps=30
>     a=rid:2 send max-fs=614400;max-fps=15
>     a=rid:3 send max-fs=230400;max-fps=30
>
> Either way is valid and sensible -- the second might be a better example,
> since it demonstrates the use of simulcast without payload-restricted RID
> lines. You don't want to give the erroneous impression that "pt" is required
> on the RID line for simulcast to work.
>
>

Agree that we should go with the later as example here. The first part 
of the comment I would refer back to the main open issue.

Cheers

Magnus Westerlund

----------------------------------------------------------------------
Network Architecture & Protocols, Ericsson Research
----------------------------------------------------------------------
Ericsson AB                 | Phone  [phone]
Torshamnsgatan 23           | Mobile [phone]
SE-164 80 Stockholm, Sweden | mailto: [e-mail]
----------------------------------------------------------------------