Re: [tsvwg] WGLC for draft-ietf-tsvwg-byte-pkt-congest-10 - to conclude 28th June 2013

[John Leslie <[e-mail]>]

Gorry Fairhurst <[e-mail]> wrote:
> 
> This email announces the beginning of a short working group last call 
> for draft-ietf-tsvwg-byte-pkt-congest-10, "Byte and Packet Congestion 
> Notification".

   First of all, I wish to applaud the authors for expanding the question
of scaling congestion control signals and responses to include queue
length in time.

   I agree with the authors that _when_ queue size cannot be measured in
time, the remaining question is whether to measure bytes or packets.

   Alas, I do not agree with the authors that measuring queue size in
_payload_ bytes is useful -- the measurement should be bytes-on-the-wire.
It is because of the difficulties in measuring byte-on-the-wire that I
tend to prefer measuring the queue length in packets.

   But arguing that becomes pointless when queue size can be measured in
time.

   Considering the current work in progress on CoDel and PIE, etc., I
wonder whether it actually makes sense to nail down the question of
what to do when we can't measure queue length in time.

   Thus, I do not favor trying to finish this draft at this time.

   If we do wish to finish this draft at this time, I have other questions
that I hope we will consider.

   I also note the Fred Baker draft revising RFC2309 being considered
in the proposed AQM WG. These two drafts overlap so much that I believe
a single co-ordinated draft revising 2309 is wiser.

--
John Leslie <[e-mail]>