Re: [tsvwg] L4S vs SCE

["Scheffenegger, Richard" <[e-mail]>]

Gorry, Jonathan,


Am 20.11.2019 um 05:23 schrieb Jonathan Morton:
>> On 20 Nov, 2019, at 12:03 pm, G Fairhurst <[e-mail]> wrote:
>>
>> Would an SCE router likely preserve or change the queueing of ECT(1) L4S traffic? - is this behvaiour different to any existing ECN-marking router?
>
> SCE middleboxes don't queue or handle ECN(1)-marked or CE-marked traffic any differently from ECT(0) traffic.  The a-priori ECT(1) marking merely removes the option of applying an SCE mark.

Will a SCE middlebox keep a rolling update of the incoming fraction of
ECT(1) vs. the fraction it would want to (re-)mark with ECT(1)? Or is
"ships-in-the-night" behavior between two fractionally congested SCE
middleboxes sufficient with an overall remarking probability of
1-((1-p1)*(1-p2))

With regular RED, ships-in-the-night was acepted, but the typical
marking ratios for 3168-TCP would be an order of magnitude lower,
limiting the deviation compared to "correct" max(p1, p2) marking.

As SCE has a target marking probability (ect0 -> ect1) from close to 50%
under steady state, such couplings between consecutive bottlenecks would
be much more prononounced.

>> And would SCE traffic arriving at the L4S queue meet L4S expectations in terms of responsiveness?
>
> Yes, since the MD response to CE is actually more severe/conservative than what L4S expects.

Well, SCE would be starved out due to the high p of CE marks in a L4S
queue, even if not sharing the Q.

[...]

> Is that sufficient, or does a more detailed analysis need to be organised?

A good analytical model of the sce cc response function would be good to
have always.