[Codel] [PATCH 1/2] codel: Controlled Delay AQM

[Jim Gettys]

On 05/07/2012 03:16 PM, Eric Dumazet wrote:
> On Mon, 2012-05-07 at 14:50 -0400, Jim Gettys wrote:
>
>> You want to turn it on in CeroWrt: sure.  Expert audience, who may be
>> able to notice when things go sour, fine....
>>
>> Neither you, nor I are expert at the current state of ECN in the global
>> Internet: but Steve Bauer is.  We can have that conversation with him
>> shortly.
>>
>> So for now "First, do no harm" should be our mantra....
>>
>> So the upstream patch default state should be off.  We don't want to mix
>> "codel doesn't work" with "ecn is busted someplace", as it gets its trials.
>>
> ecn is negociated by peers, not a qdisc.
>
> ecn at qdisc level is only marking an already enabled ECN packet.
>
> Since most packets are not ECN enabled, they are dropped instead.

That's certainly the way ECN is supposed to work right now.

>
> If you read my ECN addition to Codel, you can see its disabled by
> default already. Its a toy for early adopters.
I think it is safe for it to behave the rest of the way Linux ECN
support does right now: it only gets used if the peer requests it.

Not clear to me there needs to be/should be any option at all: the last
conversation I had with Steve Bauer was that something north of 20% of
conversations were ECN capable. Is there one for the other instances of
ECN support in Linux?  If so, it should be keyed by the same variable,
and not be a one-off for codel.

If you wanted to test ECN separately from drop with codel, then you'd
just request ECN in the conversation (by default, OS's don't normally
request ECN today, as the remaining brokenness gets sorted out).
                    - Jim