Re: [aqm] References on AQM test results and fq_nocodel

[David Lang <[e-mail]>]

On Tue, 14 Apr 2015, Toke Høiland-Jørgensen wrote:

> Simon Barber <[e-mail]> writes:
>
>> One problem with fair queueing is that it can be gamed. By opening
>> multiple flows you achieve unfair priority. Part of Codel or PIE's
>> beauty is that they are blind to the traffic, only reacting to the
>> externally visible characteristics. This stems from the question 'what
>> is a flow?'. There is no easy answer to this question, so Codel and
>> PIE intentionally avoid the question.
>
> Sure, if we knew what the One True Queue Management Scheme that always
> did the right thing was, we probably wouldn't be having this
> conversation. :)

if the fq portion is being gamed, how severe can the imbalance be? Is it a 
matter that if there are N flows without gaming the system, and each is getting 
1/N bandwith, then if a cheater uses M flows the cheater gets M/(N+M) of the 
bandwidth?

more importantly, is there any way short of a massive number of flows 
(essentially a DDOS), that this gaming of the system can hurt the latency of the 
other flows? or is it just the relative bandwidth of the different apps that 
suffers, and if an app is low bandwidth, but latency sensitive (i.e. VoIP), it 
won't be affected until it's "share" of bandwidth drops below the minimum it 
needs?

>> Of course in many practical situations some simple assumptions about
>> what a flow is do work, and in those situations fair queueing performs
>> very well. It's important to keep in mind the limitations though. Fair
>> queueing is not a panacea.
>
> Well, my focus has primarily been "why does my internet suck so much and
> what can I do to fix it". And for that (e.g. home type networks)
> fq_codel is very close to a panacea.
>
> I am well aware that there probably exists situations in which the
> fq_codel assumptions do not hold up (and we do point out a couple in the
> draft). However, I don't think I've ever seen someone actually
> *demonstrate* (you know, with data) a scenario where fq_codel performs
> worse than either Codel or PIE. If someone can point me to such a
> demonstration I'll be happy to be proved wrong... :)

Looping back to the start of the topic, the slides apparently are being read 
that fq without codel is just as good, or better than fq_codel, so codel can be 
dropped, and only fq is needed.

Is this what you are concluding from this? Or is that a misreading of something 
(the data, your work, or me misreading this thread)?

David Lang