Re: [rmcat] review of draft-ietf-avtcore-cc-feedback-message-04

[Colin Perkins <[e-mail]>]

Hi,

> On 23 Jul 2019, at 12:00, Steffen Schulze <[e-mail]> wrote:
> 
> Hi,
> 
> while reviewing draft-ietf-avtcore-cc-feedback-message-04 I found a
> minor typo you might want to fix on page 7:
> 
> then an ECN-CE mark MUST be reported that for packet;
> 
> to
> 
> then an ECN-CE mark MUST be reported for that packet;

Thanks, will fix.

> In addition I have some questions:
> 
> If a duplicate packet arrives which would update the ECN-CE field for
> the previously reported packet, as stated on page 7, how far back the
> receiver must report that?

That’s intentionally not specified, since the amount of state receivers keep, and the length of time they hold it for, varies significantly. Keeping track of duplicates for a small number of reporting intervals is likely sufficient.

> Can a CCFB packet include multiple reports for the same SSRC? (Which
> would be interesting for updating a duplicate.)

It can contain multiple reports for a single SSRC, and there can be multiple CCFB packets in a compound RTCP packet.

> The current proposal includes information about packet reception back
> to the sender for all kinds of QoS parameters, including jitter,
> packet-loss, round-trip time, etc. Therefore would it be worth
> considering to report duplicate packets to the sender as well?


I tend to think the overhead of reporting duplicates in CCFB feedback packets outweighs the utility, and we have other mechanisms to report them. It could be added if there were WG consensus to do so, of course.

Colin