On 20 Sep, 2014, at 12:03 pm, Steinar H. Gunderson wrote: > On Sat, Sep 20, 2014 at 02:33:06AM +0300, Dave Taht wrote: >> http://tools.ietf.org/html/draft-gettys-iw10-considered-harmful-00 > > The pedant in me wants to point out that 4 -> 10 is not “2.5 times worse”, > but “2.5 times as bad” or “1.5 times worse” (just as 4 -> 5 is “20% worse”, > “0.2 times worse” or “1.2 times as bad”).
ISTR seeing some concrete test data showing that IW10 doesn't even work as designed, unless TCP pacing of some type is used to spread out the burst. That's *despite* bloated buffers. Really puts the nail in the coffin, if you ask me. The recent work on SQM could be added to the list of mitigation measures. Also, by keeping inter-flow latency low, it greatly reduces the original motivation for IW10 in the first place. - Jonathan Morton _______________________________________________ Bloat mailing list [email protected] https://lists.bufferbloat.net/listinfo/bloat
