[Cerowrt-devel] [Make-wifi-fast] [Bloat] Little's Law mea culpa, but not invalidating my main point

Jonathan Morton chromatix99 at gmail.com
Mon Jul 12 17:54:00 EDT 2021

> On 12 Jul, 2021, at 11:04 pm, Bob McMahon via Make-wifi-fast <make-wifi-fast at lists.bufferbloat.net> wrote:
> "Flow control in store-and-forward computer networks is appropriate for decentralized execution. A formal description of a class of "decentralized flow control algorithms" is given. The feasibility of maximizing power with such algorithms is investigated. On the assumption that communication links behave like M/M/1 servers it is shown that no "decentralized flow control algorithm" can maximize network power. Power has been suggested in the literature as a network performance objective. It is also shown that no objective based only on the users' throughputs and average delay is decentralizable. Finally, a restricted class of algorithms cannot even approximate power."
> https://ieeexplore.ieee.org/document/1095152
> Did Jaffe make a mistake?

I would suggest that if you model traffic as having no control feedback, you will inevitably find that no control occurs.  But real Internet traffic *does* have control feedback - though it was introduced some time *after* Jaffe's paper, so we can forgive him for a degree of ignorance on that point.  Perhaps Jaffe effectively predicted the ARPANET congestion collapse events with his analysis.

> Also, it's been observed that latency is non-parametric in it's distributions and computing gaussians per the central limit theorem for OWD feedback loops aren't effective. How does one design a control loop around things that are non-parametric? It also begs the question, what are the feed forward knobs that can actually help?

Control at endpoints benefits greatly from even small amounts of information supplied by the network about the degree of congestion present on the path.  This is the role played first by packets lost at queue overflow, then deliberately dropped by AQMs, then marked using the ECN mechanism rather than dropped.

AQM algorithms can be exceedingly simple, or they can be rather sophisticated.  Increased levels of sophistication in both the AQM and the endpoint's congestion control algorithm may be used to increase the "network power" actually obtained.  The required level of complexity for each, achieving reasonably good results, is however quite low.

 - Jonathan Morton

More information about the Cerowrt-devel mailing list