Re: [Cake] cake for net-next 4.8

[Dave Taht <dave.taht@gmail.com>]

On Tue, Sep 27, 2016 at 10:52 AM, Jonathan Morton <chromatix99@gmail.com> wrote:
>
>> On 25 Sep, 2016, at 21:30, Dave Taht <dave.taht@gmail.com> wrote:
>>
>> Judging from me tearing apart how TCP BBR works (presently) with ecn,
>> it looks like we need to add the equivalent to fq_codel ce_threshold
>> behaviors as well.
>
> If I’m reading the legend correctly, you are setting ce_threshold to 1ms to get the better-controlled result.  But that effectively disables the codel algorithm and turns it into a simple “mark all packets over 1ms sojourn” for ECN capable traffic, because it’s a tighter limit than codel’s target.  That’s too aggressive for non-BBR traffic.

Yes it is. :) However the consensus appears to be that ECN should be
an earlier signal than drop, and the work over on the tcp-prague list
centers around repurposing ECT(1) as more like a DCTCP multi-bit
signal.

ce_threshold 1ms is more like signaling loudly - "this is a solid
indicator of  your real OWD" and "you should fiddle with your TCP's
'gain' to compensate for it" - and it's certainly a lot simpler than
codel to do it this way. I haven't got around to fully evaluating a
comparison between BBR with ce_threshold and a non-ecn enabled TCP vs
codel, concurrently (or implementing different behaviors for ECT(1) in
the TCP negotiation step) ...

and... I'm really not sure if what I've seen with ce_threshold is the
desired behavior, vs a vs BBR, thus far - but I'd like to see the
option for it enter cake.

Also, in doing in my network this way, it became apparent to me that -
like several ecn encapsulation rfcs suggest - that when a packet is
already marked CE, and we want to also mark it CE, that the rightest
answer is to drop the packet - not giving pre-CE-marked flows a free
ride.

> In these cases, I think you have to relax and let the FQ action take care of it.

And yes, FQ handles it nicely. Got pics of that somewhere. I did do a
test with ce_threshold + ecn + bbr, vs no ecn and cubic, with the
usual lovely fq'd result.

>  - Jonathan Morton
>



-- 
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org