From owner-freebsd-net Sat Apr 6 18:45:30 2002 Delivered-To: freebsd-net@freebsd.org Received: from rwcrmhc54.attbi.com (rwcrmhc54.attbi.com [216.148.227.87]) by hub.freebsd.org (Postfix) with ESMTP id 07D5A37B405 for ; Sat, 6 Apr 2002 18:45:28 -0800 (PST) Received: from blossom.cjclark.org ([12.234.91.48]) by rwcrmhc54.attbi.com (InterMail vM.4.01.03.27 201-229-121-127-20010626) with ESMTP id <20020407024527.GOVO15826.rwcrmhc54.attbi.com@blossom.cjclark.org>; Sun, 7 Apr 2002 02:45:27 +0000 Received: (from cjc@localhost) by blossom.cjclark.org (8.11.6/8.11.6) id g372jQE71070; Sat, 6 Apr 2002 18:45:26 -0800 (PST) (envelope-from cjc) Date: Sat, 6 Apr 2002 18:45:26 -0800 From: "Crist J. Clark" To: Luigi Rizzo Cc: Andreas Klemm , freebsd-net@FreeBSD.ORG Subject: Re: better DSL bandwidth usage by priorizing ACKs in outgoing packets over others Message-ID: <20020406184526.E70207@blossom.cjclark.org> References: <20020402163912.GD4307@titan.klemm.gtn.com> <20020406065735.A29144@iguana.icir.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <20020406065735.A29144@iguana.icir.org>; from luigi@iet.unipi.it on Sat, Apr 06, 2002 at 06:57:36AM -0800 X-URL: http://people.freebsd.org/~cjc/ Sender: owner-freebsd-net@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org On Sat, Apr 06, 2002 at 06:57:36AM -0800, Luigi Rizzo wrote: > On Tue, Apr 02, 2002 at 06:39:12PM +0200, Andreas Klemm wrote: > > we would need a minor tweak to the ipfw code so that it can match > packets whose size is less than X bytes (so the mechanism is general > enough to be used for other things). This could be done in a matter > of 1hour or less, most of the time would be wasted in figuring out > a way to implement it that does not break binary compatibility. Actually, could just match ACK-only packets, no PSH? ...I'm getting some severe deja vu with this topic. But I can't recall what the exact subject was previously. > Once we have done this, we can define a dummynet pipe with > a bandwidth <= the bottleneck (128kbit/s), and use"queue" rules > to privilege the acks wrt other traffic. > > Something like > > > ipfw pipe 10 config bw 100kbit/s > ipfw queue 1 config weight 1 pipe 10 > ipfw queue 100 config weight 100 pipe 10 > ipfw add queue 100 tcp from any to ${outside} shorter-than 80 > ipfw add queue 1 ip from any to ${outside} > > This said, i have never seen terribly bad effects when cvsupping > and doing other things. If there is something which goes to its > knees, this is the disk. On a previous Internet provider, I had silent PMTU issues somewhere downstream. Ploss went through the roof when you got above 1000 bytes-per-packet upstream. -- Crist J. Clark | cjclark@alum.mit.edu | cjclark@jhu.edu http://people.freebsd.org/~cjc/ | cjc@freebsd.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message