[Bloat] ultrafast broadband conference june 27-30

Dave Taht dave.taht at gmail.com
Sat Jun 18 10:07:59 EDT 2016


On Sat, Jun 18, 2016 at 6:55 AM, Jonathan Morton <chromatix99 at gmail.com> wrote:
>
>> On 18 Jun, 2016, at 14:22, Jan Ceuleers <jan.ceuleers at gmail.com> wrote:
>>
>> Would it be worth pursuing an RFC on how to properly test the
>> performance of an internet access circuit and/or of a path across an IP
>> network? RFC compliance might be a tool to get Ookla to take notice.

My answer has become to try to "get in the face" of more existing
standards bodies working at these layers.

Before that, it was "find a way to sit down with the actual driver
developer to go make these *very simple* changes to the underlying
firmware blob". A BQL-like technique is *not rocket science*. My
estimate was less than 3 days worth of engineering for someone
familiar with the firmware source, coupled with someone familiar with
the queue theory.

Merely limiting the maximum number of packets queued in the firmware
to 2x what is needed for interleave at that rate to work, is even
simpler, if less effective.

> I think that could happen, if someone sits down and drafts one.  The “AQM Characterization Guidelines” one is coming along nicely, and is obviously related.
>
>  - Jonathan Morton
>
> _______________________________________________
> Bloat mailing list
> Bloat at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat



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



More information about the Bloat mailing list