[AVT] Comment on draft-ietf-avt-smpte292-video-08.txt

[Stephen Casner <casner@acm.org>]

This comment was triggered by an email from Chuck Harrison, which was
intern triggerd by one of my questions in the RTP spec/profile series.
Please take these comments as arriving under the IESG Last Call.

draft-ietf-avt-smpte292-video-08.txt includes the following paragraph
in its Section 6:

   RFC1889 recommends transmission of RTCP packets every 5 seconds or at a
   reduced minimum in seconds of 360 divided by the session bandwidth in
   kilobits/second. At 1.485 Gbps the reduced minimum interval computes to
   0.2ms or 4028 packets per second.

I have three comments on this:

  - I'm not sure why I did not trip on this before, but RFC 1889 does
    _not_ recommend transmission of RTCP packets every 5 seconds.  It
    recommends RTCP transmission based on a scalable timer with 50%
    randomization and a minimum _average_ interval of 5 seconds.  It
    is way too tempting already for implementers to put in a fixed
    5-second timer.  Let's not reinforce that error.

  - RFC 1889 does _not_ include the scaling of the minimum interval.
    That was added in draft-ietf-avt-rtp-new.  If smpte292-video gets
    to RFC first, well...

  - Maybe it would appropriate to add another sentence saying that
    4028 is likely to be excessive for most applications and perhaps
    suggesting another value.  The rate only needs to be fast enough
    to track clock skew and to provide enough packets during the
    23-second octet-count wrap interval to reliably track the wraps.

                                                        -- Steve

_______________________________________________
Audio/Video Transport Working Group
avt@ietf.org
https://www1.ietf.org/mailman/listinfo/avt