[Ecn-sane] [tsvwg] per-flow scheduling

Jonathan Morton chromatix99 at gmail.com
Mon Jun 24 16:14:06 EDT 2019


> On 24 Jun, 2019, at 10:50 pm, David P. Reed <dpreed at deepplum.com> wrote:
> 
> My overall point here is that you seem to live in a world of academic-like purity - all TCP connections are essentially huge file transfers, where there are no delays on production or consumption of packets at the endpoint, there is no multiplexing or scheduling of processes in the endpoint operating systems, etc.

On the contrary, I've found that per-flow queuing algorithms like DRR++ cope naturally and very nicely with all sorts of deviations from the ideal, including for example the wild variations in goodput and RTT associated with wifi links.

These are exactly the kinds of complication that I imagine - and not merely in the abstract but through observation - that a pure end-to-end approach would have great difficulty in accommodating elegantly.

 - Jonathan Morton



More information about the Ecn-sane mailing list