[Cerowrt-devel] TSO sizing and FQ scheduler

[Cerowrt-devel] TSO sizing and FQ scheduler

[Maciej Soltysiak]

Hi list,

3.12 landed with TSO sizing and FQ scheduler. Is there significant
benefit of trying to port these to Cero's 3.10 ?

I'm assuming we're not going head on to 3.12 for kernel base for cero as of yet?

Best regards,
Maciej

Re: [Cerowrt-devel] TSO sizing and FQ scheduler

[Sebastian Moeller]

Hi Maciej,



On Nov 5, 2013, at 14:22 , Maciej Soltysiak <maciej@soltysiak.com> wrote:

> Hi list,
> 
> 3.12 landed with TSO sizing and FQ scheduler. Is there significant
> benefit of trying to port these to Cero's 3.10 ?

	According to Eric Dumazet, these two help for flows terminating on the device in question, not for flows just passing through the device. So unless your cerowrt router offers lots of network services it most likely will not profit from these features… Also, IIRC, we disable TSO on cerowrt by default (though TSO sizing might mean that this decision could be revisited).

> 
> I'm assuming we're not going head on to 3.12 for kernel base for cero as of yet?

	I would hope for 3.10 to be the kernel for the forceable future, due to its promised 2 years? maintenance window.


Best
	Sebastian

> 
> Best regards,
> Maciej
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel

Re: [Cerowrt-devel] TSO sizing and FQ scheduler

[Dave Taht]

On Tue, Nov 05, 2013 at 02:39:28PM +0100, Sebastian Moeller wrote:
> Hi Maciej,
> 
> 
> 
> On Nov 5, 2013, at 14:22 , Maciej Soltysiak <maciej@soltysiak.com> wrote:
> 
> > Hi list,
> > 
> > 3.12 landed with TSO sizing and FQ scheduler. Is there significant
> > benefit of trying to port these to Cero's 3.10 ?
> 
> 	According to Eric Dumazet, these two help for flows terminating on the device in question, not for flows just passing through the device. So unless your cerowrt router offers lots of network services it most likely will not profit from these features… Also, IIRC, we disable TSO on cerowrt by default (though TSO sizing might mean that this decision could be revisited).
> 
> > 

Of interest in backporting FQ is to be able to analyze its performance
on routed streams vs fq_codel. It uses a finer grained FQ, on a 4 tuple, 
but does not do queue management, so I expect the results will end up looking
like SFQ or QFQ with bigger buffers - good fairness but catastrophic effects on 
fat streams, and a lot more collisions.

This is a slide I didn't get to show at ietf, showing what SFQ + longest
queue drop looks like.

http://snapon.lab.bufferbloat.net/~d/results_3.12/3.12/ecn/rtt_fair_download=5500-upload=20000-noecn-sfq-target=5ms-delay=240ms-2.svg

vs fq_codel

http://snapon.lab.bufferbloat.net/~d/results_3.12/3.12/ecn/rtt_fair_download=5500-upload=20000-ecn-fq_codel-target=5ms-delay=240ms-2.svg

vs pie

http://snapon.lab.bufferbloat.net/~d/results_3.12/3.12/ecn/rtt_fair_download=5500-upload=20000-ecn-pie-target=20ms-delay=240ms-2.svg

I expect the FQ scheduler to crash a teeny router with its defaults also...

Looking at the cpu usage would be interesting too...

I'm tempted to backport it, I am! I am!


> > I'm assuming we're not going head on to 3.12 for kernel base for cero as of yet?
> 
> 	I would hope for 3.10 to be the kernel for the forceable future, due to its promised 2 years? maintenance window.

I intend to freeze on some variant of 3.10 for quite some time, yes.

At the moment I don't really have time to backport the new stuff
to cerowrt, I am starting a new job (yea!), and just want to make cero stable
ASAP so I can ramp up on that. 

> 
> 
> Best
> 	Sebastian
> 
> > 
> > Best regards,
> > Maciej
> > _______________________________________________
> > Cerowrt-devel mailing list
> > Cerowrt-devel@lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/cerowrt-devel
> 
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel