Re: [AVTCORE] WGLC review on draft-ietf-avtcore-multi-media-rtp-session-09

[Colin Perkins <[e-mail]>]

On 10 Sep 2015, at 17:55, Asveren, Tolga <[e-mail]> wrote:
>>> The RTCP timing rules in RFC 3550 don't support different reporting 
>>> intervals for different SSRCs (other than a sender/receiver split), 
>>> and doing so runs a risk of causing inappropriate RTCP timeouts. We 
>>> could add a reference to RFC
>>> 3550 here, if it helps?
>> [TOLGA] Can't this be changed? On a high level, aren't we talking about two different RTCP streams here? Why should their timing be aligned in any way?
> 
> Their timing isn’t aligned, they merely use the same base RTCP reporting interval, before randomisation, and modulo the difference between sender and receiver bandwidth allocation. This is the usual RFC 3550 RTCP timing rules. If you want different reporting intervals, you need to use separate RTP sessions.
> [TOLGA] I was also referring to reporting timing but admittedly not being clear. But then again, why should their timing be the same? Is there a technical justification or "just" because RFC3550 (which probably did not consider the use case defined/explained in this draft as it was not defined yet) says so?

The technical justification is as I said above: you get inappropriate timeouts, because the algorithm is based on all SSRCs using the same base reporting interval. 

Colin





> Thanks,
> Tolga
> 
>> -----Original Message-----
>> From: Colin Perkins [mailto:[e-mail]]
>> Sent: Thursday, September 10, 2015 6:53 AM
>> To: Asveren, Tolga <[e-mail]>
>> Cc: [e-mail]; Ali Begen <[e-mail]>
>> Subject: Re: [AVTCORE] WGLC review on draft-ietf-avtcore-multi-media-rtp-
>> session-09
>> 
>>>> On 10 Sep 2015, at 11:27, Asveren, Tolga <[e-mail]>
>>> wrote:
>>> 
>>> Please see inline for questions/answers/comments.
>>> 
>>> Thanks,
>>> Tolga
>>> 
>>>> -----Original Message-----
>>>> From: Colin Perkins [mailto:[e-mail]]
>>>> Sent: Monday, September 07, 2015 7:35 AM
>>>> To: Asveren, Tolga <[e-mail]>
>>>> Cc: [e-mail]; Ali Begen <[e-mail]>
>>>> Subject: Re: [AVTCORE] WGLC review on
>>>> draft-ietf-avtcore-multi-media-rtp-
>>>> session-09
>>>> 
>>>> Hi,
>>>> 
>>>> Thanks for the review; some responses inline.
>>>> 
>>>> On 6 Sep 2015, at 14:24, Asveren, Tolga <[e-mail]>
>> wrote:
>>>>> I did not monitor the progress of this draft over its lifetime, so
>>>>> this is more of an "outsider's review" (which probably is useful in
>>>>> its own right)
>>>>> 
>>>>> i- 2. Terminology
>>>>> "Media Type:  The general type of media data used by a real-time
>>>>>    application.  The media type corresponds to the value used in the
>>>>>    <media> field of an SDP m= line.  The media types defined at the
>>>>>    time of this writing are "audio", "video", "text", "application",
>>>>>    and "message"."
>>>>> 
>>>>> - It could be good to provide a reference to the corresponding IANA
>> registry.
>>>>> - I know it is not very common but still defined/possibly used, is there a
>> specific reason not to include “image” (e.g., image/t38, image/g3fax) in this
>> list?
>>>> 
>>>> I copied the list from draft-ietf-mmusic-rfc4566bis-15, which doesn't
>>>> mention "image". I agree we should mention it, as should 4566bis.
>>>> (I've cc'd Ali, for comment).
>>>> 
>>>>> ii- 3.  Background and Motivation
>>>>> In general, this section seems trying to do a “hard sell", which IMHO can
>> be softened up a little bit.
>>>>> 
>>>>> - "However, we note that many RTP-using application don't use
>>>>> network  QoS features, and don't expect or desire any separation in
>>>>> network  treatment of their media packets, independent of whether
>>>>> they are  audio, video or text.  When an application has no such
>>>>> desire, it  doesn't need to provide a transport flow structure that
>>>>> simplifies  flow based QoS."
>>>>> This really depends on the environment and deployment model. In
>>>>> addition to that, providing QoS with DiffServ is still possible when
>>>>> multiple types of media are sent in a single RTP session as
>>>>> explained in draft-ietf-dart-dscp-rtp (which is already referenced
>>>>> by this
>>>>> draft)
>>>> 
>>>> It's not clear what change you want here. Do you have a suggestion
>>>> for how to update the text?
>>> [TOLGA] Maybe just taking out QoS related part?
>> 
>> Since RFC 3550 explicitly calls out QoS as a reason for sending multiple media
>> streams on different transport layer flows, I don’t think we can do this.
>> 
>>>>> -  "1.  increased delay to establish a complete session, since each of
>>>>>     the transport layer flows needs to be negotiated and established;"
>>>>> At least negotiation part can happen in parallel. And when multiple
>> streams use the same RTP session, isn’t there still a need to negotiate
>> different SSRC values, one per stream? A reference to RFC5576 could be
>> useful as well.
>>>> 
>>>> Negotiation can only happen in parallel to a certain extent, before
>>>> the amount of STUN traffic causes congestion. The SSRCs also don't
>>>> necessarily have to be negotiated in advance for either model.
>>> [TOLGA] It could be useful to provide more information about how
>> demultiplexing based on SSRC would work when multiple streams are sent in
>> the same RTP session. Isn't there inherently a need to associate a stream
>> with a particular SSRC? How is this done? Just a description of that, maybe in
>> Section 5.2. so that the logic is obvious to people (particularly to first-time
>> readers).
>> 
>> RTP doesn’t have a requirement to associate a stream with an SSRC. Certain
>> classes of application might have such a requirement, but that’s a signalling
>> issue, and so not something for this draft (it’s something for BUNDLE, or
>> similar, to specify).
>> 
>>>>> iii- 5.3.  Per-SSRC Media Type Restrictions This section provides the
>> justification why different SSRC but be used for audio and video streams but
>> not really why a SSRC can’t be re-used for streams of different type. It would
>> be good to cover that aspect as well.
>>>> 
>>>> Sorry, I'm not sure what you're asking here. Can you rephrase?
>>> [TOLGA] Mea culpa, I think 5.3 already explains why the same SSRC can't be
>> re-used for streams of different type:
>>> "The payload type will inform a receiver which
>>>  media type the SSRC is being used for.  Thus the payload type MUST be
>>>  unique across all of the payload configurations independent of media
>>>  type that is used in the RTP session."
>>> I guess this section provides the justification. Please correct me if I am
>> wrong.
>> 
>> The 2nd and 3rd paragraphs of Section 5.3 explain why an SSRC can’t
>> simultaneously be used for audio and video streams.
>> 
>> An SSRC cannot start using audio and then switch to using video, since that
>> would require a change in RTP timestamp rate (which RFC 7160 explains
>> doesn’t work). However, an SSRC can send audio then leave the session (by
>> sending an RTCP BYE, or stopping sending RTP and RTCP for long enough that
>> it times out), then a new source can start with the same SSRC number
>> sending video; this doesn’t cause confusion because the audio SSRC has
>> gone before the video SSRC starts.
>> 
>> 
>>>>> iv- 4.  Applicability
>>>>> "Compatible RTCP Behaviour:  The RTCP timing rules enforce a single
>>>>>    RTCP reporting interval for all participants in an RTP session.
>>>>>    Flows with very different media sending rate or RTCP feedback
>>>>>    requirements cannot be multiplexed together, since this leads to
>>>>>    either excessive or insufficient RTCP for some flows, depending
>>>>>    how the RTCP session bandwidth, and hence reporting interval, is
>>>>>    configured."
>>>>> This part IMHO requires more justification. Why is it not possible to send
>> RTCP in different rates for different flows? Some RTCP packets could have
>> reports pertaining to a single stream whereas the other ones (when timing
>> for reports for different streams overlaps or is sufficiently close enough) for
>> multiple streams. Aggregate RTCP bandwidth need shouldn’t be a
>> constraining factor just because a single transport flow is used AFAICS. If that
>> is not the case, it really could be useful to explain why.
>>>> 
>>>> The RTCP timing rules in RFC 3550 don't support different reporting
>>>> intervals for different SSRCs (other than a sender/receiver split),
>>>> and doing so runs a risk of causing inappropriate RTCP timeouts. We
>>>> could add a reference to RFC
>>>> 3550 here, if it helps?
>>> [TOLGA] Can't this be changed? On a high level, aren't we talking about two
>> different RTCP streams here? Why should their timing be aligned in any way?
>> 
>> Their timing isn’t aligned, they merely use the same base RTCP reporting
>> interval, before randomisation, and modulo the difference between sender
>> and receiver bandwidth allocation. This is the usual RFC 3550 RTCP timing
>> rules. If you want different reporting intervals, you need to use separate RTP
>> sessions.
>> 
>>>>> v- 7.  Signalling
>>>>> "When using SDP signalling, the BUNDLE extension
>>>>> [I-D.ietf-mmusic-sdp-bundle-negotiation] is used to signal RTP
>>>>> sessions containing multiple media types."
>>>>> It could be a good idea to reference relevant sections as the
>>>>> bundle-draft
>>>> has some irrelevant aspects (arguably which even could be confusing
>>>> in the context of this specification).
>>>> 
>>>> Can you be specific on how BUNDLE should be referenced?
>>> [TOLGA] Maybe just specifying which sections are relevant and which are
>> not, e.g. "14. RTP/RTCP extensions for identification-tag transport" in draft-
>> ietf-mmusic-sdp-bundle-negotiation-23.txt, is this part relevant to  draft-ietf-
>> avtcore-multi-media-rtp-session-09?
>> 
>> This is to address the stream identification issue you raise above.
>> 
>> --
>> Colin Perkins
>> https://csperkins.org/
>