Re: [AVTCORE] [rtcweb] [tsvwg] WG Last Call on changes: draft-ietf-avtcore-rtp-circuit-breakers-16

[John Leslie <[e-mail]>]

Black, David <[e-mail]> wrote:
> 
> In the new ECN text in Section 7, I see:
> 
>    Given the above issues, implementations MAY ignore ECN-CE marks when	
>    determining if the congestion circuit breaker triggers, since	
>    excessive persistent congestion will eventually lead to packet loss	
>...
> I think "MAY" is inappropriate here - this text reads like a "SHOULD NOT"
> requirement with an explanation of what happens when something else
> is done.

   I'm very sympathetic to your issue: there are implementations of ECN
out there which _will_not_ drop an ECN-capable packet before the sending
queue has overflowed.

   But at the same time, we're headed into changes where ECN-CE marking
will become more frequent than packet-drop.

   We must find a way to avoid ECN-CE triggering the congestion circuit
breaker when the forwarding node has not reached a level of congestion
which could trigger packet loss.

   I'd prefer language (perhaps in an RFC-3168-bis) stating that packet
drop MUST me used, at least occasionally, on packets which would be
ECN-CE marked, when the overall congestion at that node reaches the
drop level for non-ECN-capable packets. (The forwarding node is the only
node with sufficient information to see the issue.)

   (I recommend folks interested in circuit-breaker pay attention to the
L4S BoF being scheduled for Berlin.)

--
John Leslie <[e-mail]>