I was very interested to see this draft discusses a problem with AQMs
AQM schemes like CoDel and PIE use congestion notifications to
constrain the queuing delays experienced by packets, rather than in
response to impending or actual bottleneck buffer exhaustion. With
current default delay targets, CoDel and PIE both effectively emulate
a shallow buffered bottleneck (section II, [ABE2015
<https://tools.ietf.org/html/draft-khademi-alternativebackoff-ecn-01#ref-ABE2015>]).
This
interacts acceptably for TCP connections over low BDP paths, or
highly multiplexed scenarios (lmany concurrent TCP connections).
However, it interacts badly with lightly-multiplexed cases (few
concurrent connections) over high BDP paths. Conventional TCP
backoff in such cases leads to gaps in packet transmission and
underutilisation of the path.
I think it wold be good to add some discussion of this effect to the
draft on evaluating AQM algorithms. In many access network scenarios the
paths will be lightly loaded, and sometimes higher BDPs will be
experienced. In these cases it's good to know that the AQM is not
hurting your experience.
Simon
_______________________________________________
aqm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/aqm
