Date: Sun, 27 Sep 1998 13:52:46 -0400 (EDT) From: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> To: "Allen Smith" <easmith@beatrice.rutgers.edu> Cc: ark@eltex.ru, luigi@labinfo.iet.unipi.it, kev@lab321.ru, mike@smith.net.au, net@FreeBSD.ORG Subject: Re: Packet/traffic shapper ? Message-ID: <199809271752.NAA04318@khavrinen.lcs.mit.edu> In-Reply-To: <9809260207.ZM14494@beatrice.rutgers.edu> References: <199809210951.NAA32644@paranoid.eltex.spb.ru> <9809260207.ZM14494@beatrice.rutgers.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
<<On Sat, 26 Sep 1998 02:07:58 -0400, "Allen Smith" <easmith@beatrice.rutgers.edu> said: > Not to get back into the debate regarding ALTQ's "ugliness", the > primary thing I was looking at ALTQ for was the RED (Random Early > Detection) capability of ALTQ, so that I can get the lower-priority > TCP streams to drop back their bandwidth when they're getting too > much. I once did a RED implementation for FreeBSD, mostly aping the sample code in the original RED paper. It's not all that hard, but does involve a lot of fiddling to deal with all the code that thinks it knows how interfaces should operate better than you do. One of the particular difficulties -- if you want RED to work like it's supposed to -- is to squash the internal queueing that a lot of network interface drivers do. I didn't solve that problem (``left as an exercise...''), but it should be obvious that for correct operation, it's essential for RED to be able to determine the actual instantaneous length of the queue. -GAWollman -- Garrett A. Wollman | O Siem / We are all family / O Siem / We're all the same wollman@lcs.mit.edu | O Siem / The fires of freedom Opinions not those of| Dance in the burning flame MIT, LCS, CRS, or NSA| - Susan Aglukark and Chad Irschick 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?199809271752.NAA04318>