[Iccrg] Why don't we stop treating ECN and loss similarly?
John Leslie
john at jlc.net
Sat Oct 27 12:55:59 BST 2012
Michael Welzl <michawe at ifi.uio.no> wrote:
>
> According to RFC 3168, senders must react to ECN just as if packets
> had been dropped. This is to maintain fairness between ECN-compatible
> and non-compatible flows.
> Because of this requirement, AQMs cannot ECN-mark packets more
> aggressively than it drops packets from non-ECN-capable flows - else
> ECN-marked flows would be at a disadvantage.
As Bob has pointed out, theory clearly shows that the less-aggressive
reaction-to-loss will starve the more-aggressive...
> We have seen various non-standard congestion control behaviors can
> co-exist reasonably well with standard TCP in practice.
Yes! At the risk of setting-off Bob, this suggests that the theory
may not completely correspond to reality. ;^)
> If it was possible to have a milder congestion reaction to ECN-based
> reaction, it would also be possible to ECN-mark packets earlier,
> leading to a bigger advantage for everyone using ECN.
This, of course, would be trivial if we had a gradation of ECN
marking...
> And none of this is possible when we have the "treat an ECN mark
> just like loss" rule in place.
Rules can be relaxed in research projects...
> Hence, my question: to incentivize ECN usage and enable better
> behavior when it's used, shouldn't we remove this rule?
There is certainly room for research which "partially" relaxes
this rule.
> Note that this is not even about a more fine-grain interpretation of
> ECN feedback - it's more like an intermediate step.
I'll remind folks that the RMCAT WG needs to convince folks that
a less-aggressive response to ECN/loss, coupled with a less-
aggressive ramp-up in the absence of ECN/loss, is safe.
Early indications are that this will prove acceptable...
--
John Leslie <john at jlc.net>
More information about the Iccrg
mailing list