Date: Thu, 8 Aug 2002 19:10:56 -0500 (CDT) From: Mike Silbersack <silby@silby.com> To: "Henderson, Thomas R" <thomas.r.henderson@boeing.com> Cc: "'freebsd-net@freebsd.org'" <freebsd-net@freebsd.org> Subject: RE: [patch] Possible newreno fix, please test Message-ID: <20020808185959.Q52867-100000@patrocles.silby.com> In-Reply-To: <00EBC850E752CC46B8509DAB4D0D2CB910692C@xch-nw-29.nw.nos.boeing.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 8 Aug 2002, Henderson, Thomas R wrote: > A couple of years ago I helped Niels Provos port our Berkeley Daedalus > project SACK/FACK/NewReno code (for BSDi) to OpenBSD. I did contribute the > newreno piece originally to the Berkeley code. > (ftp://daedalus.cs.berkeley.edu/pub/tcpsack/bsdi-3.0) Someone ported that > newreno code into freebsd (looks like 1.107.2.6 on tcp_input.c-- jlemon) but > not the sack code. I understand that. However, you're one of the attributed people in the OpenBSD commit message which is apparently of a later vintage. I'm curious what was submitted as part of that change. > My question was due to the fact that the proposed patch diverged from the > NewReno RFC and Berkeley code-- I was curious to find out why and what > problems people found to motivate that patch. > > Tom There have been a few solid reports of newreno actually decreasing performance greatly in some network environments. Matt Dillon helped reduce this somewhat by removing the maxburst limit, but some other gremlins still seem to inhabit the newreno code. I posted this patch because I found it in the OpenBSD repository, with your name among those credited for the changes. I was unable to find any messages on openbsd mailing lists talking about the patch in question, so I'm short on information here. Personally, I don't believe that the code in question was correct in the first place. Isn't the congestion window supposed to be increased for every dupack received, not just every third? In short, I'm not sure that the patch is entirely correct, but I wanted to see if it provided any noticeable improvements for those who had complained of poor performance in the past. Unfortunately, none of those people have responded yet. Mike "Silby" Silbersack To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020808185959.Q52867-100000>