Re: [tsvwg] new tests of L4S RTT fairness and intra-flow latency

[Sebastian Moeller <[e-mail]>]

Hi Alex,

more below in-line, pre-fixed [SM].

> On Nov 15, 2020, at 14:06, [e-mail] wrote:
> 
> Hi Pete,
>  Interesting data. 
> Maybe I've missed it, but I don't see whether these tests were run against a version of TCP-Prague having the RTT-independence workaround

	[SM] Which "RTT-independence workaround" are you talking about here? All that is implemented in TCP Prague is a crutch around the issue that DualPI2 catastrophically fails at its core job, with short RTT. That is a L4S/DualPi2-unique problem that is orthogonal to reduction of RTT-bias as mentioned in the Prague requirements. Just because Koen and Bob, framed this as addressing "RTT bias" does not make that assertion a fact...

> mentioned during interim-2020-tsvwg-01 [1], and if so what its mode was. Can you clarify that?

	The tests are well documented and use the latest Linux kernel from team L4S' repository https://github.com/L4STeam/linux/tree/testing (which happens to be the most recently changed branch as well as the default branch). This therefore refelects team L4S recommended TCP Prague configuration.

Best Regards
	Sebastian


> 
> Thanks,
> Alex
> 
> [1] Slide 16 of https://datatracker.ietf.org/meeting/interim-2020-tsvwg-01/materials/slides-interim-2020-tsvwg-01-sessa-l4s-tcp-prague-update-00.pdf
> 
> 
> 
> On Thursday, November 12, 2020, 9:03:17 PM GMT, Pete Heist <[e-mail]> wrote: 
> 
> 
> 
> Hi,
> 
> We have posted some new tests of L4S here:
> 
> https://github.com/heistp/l4s-tests
> 
> These tests cover two-flow fairness in several qdiscs with different
> path RTTs per-flow, and transient behavior in fq_codel queues.
> 
> The results raise some concerns about a bias in favor of L4S flows in
> DualPI2 queues, throughput imbalances between flows with different path
> RTTs, and intra-flow latency spikes upon rate reductions in fq_codel.
> The repo above contains a walk-through of the key findings, and links
> to more results in the Appendix.
> 
> Regards,
> Pete
> 
>