Re: [Cake] [Codel] Proposing COBALT

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

On Tue, Jun 28, 2016 at 1:40 AM, Kevin Darbyshire-Bryant
<kevin@darbyshire-bryant.me.uk> wrote:
>
>
> On 28/06/16 03:51, Jonathan Morton wrote:
>>>
>>> On 27 Jun, 2016, at 18:18, Kevin Darbyshire-Bryant
>>> <kevin@darbyshire-bryant.me.uk> wrote:
>>>
>>> How do you feel about switching that package to the cobalt variant for
>>> wider stress testing?
>>
>> I think the best way to do that would be to merge the cobalt branch to
>> master, but retaining it for further development.  It’s stable enough for
>> that.
>>
>> Except for the stats thing.  Best to fix that first.
>
> Would you like me to split out 'sparse_flows' and 'decaying_flows'? We have
> the space to do so, actually I've done it but won't push unless you say go.
>
> I can hear Dave groaning about stats overload from here:  To allay fears,
> the decaying_flows count is maintained as part of COBALT anyway so there's
> no extra overhead involved.  It nicely splits out the 'codel' flows from the
> 'blue' flows.

My "rule" is basically to keep no on-going (main loop) stats that are
not required for the correct functioning of the algorithm. It's
totally ok to calculate a complex stat when queried for, but there
are(were?) multiple stats in cake that are only used by a small subset
of humanity, and even those, infrequently. I'd have ripped them out
(or made them for debugging only) a long time ago if I'd got consensus
- or folk were trying to optimize again for speed. A good algorithm is
elegant and needs minimal state in the first place...

If you have a stat you need for the algorithm to work, by all means
report on it.

> Kevin
>
>>
>>   - Jonathan Morton
>>
>
> _______________________________________________
> Cake mailing list
> Cake@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cake



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