From owner-freebsd-hackers Mon Nov 22 16:44:14 1999 Delivered-To: freebsd-hackers@freebsd.org Received: from alpo.whistle.com (alpo.whistle.com [207.76.204.38]) by hub.freebsd.org (Postfix) with ESMTP id D53F614CA5 for ; Mon, 22 Nov 1999 16:44:11 -0800 (PST) (envelope-from julian@whistle.com) Received: from current1.whistle.com (current1.whistle.com [207.76.205.22]) by alpo.whistle.com (8.9.1a/8.9.1) with SMTP id QAA31108; Mon, 22 Nov 1999 16:42:29 -0800 (PST) Message-ID: <3839E2F5.2781E494@whistle.com> Date: Mon, 22 Nov 1999 16:42:29 -0800 From: Julian Elischer Organization: Whistle Communications X-Mailer: Mozilla 3.04Gold (X11; I; FreeBSD 3.3-RELEASE i386) MIME-Version: 1.0 To: David Gilbert Cc: Josef Karthauser , freebsd-hackers@FreeBSD.ORG Subject: Re: PPPoE offer. References: <14322.15597.505566.676863@trooper.velocet.ca> <19990929211210.H86792@florence.pavilion.net> <14322.30121.178080.621999@trooper.velocet.ca> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG David Gilbert wrote: > > >>>>> "Josef" == Josef Karthauser writes: > > Josef> As far as I'm aware it _does_ work - in the form of user-ppp > Josef> (/usr/sbin/ppp), maintained by Brian. Why do you need to use > Josef> kernel ppp - it's a mess :) > > In some discussions with some local BSD hackers, many claimed that I > would never get the performance I required out of user-ppp. The basic > requirement is that we handle somewhere between 5k and 10K > connections (on some amount of CPU). Were would I find recent patches > to user-ppp to receive PPPoE streams? Now in 3.3-Stable and -current. We also have patches in the works to allow full kernel bypass of the ppp daemon in the common case, resulting in Mucho performance increase. with the version we have now, a 100KB/sec ppp session took 6% of a P6-200 (the other end was a 486DX50 :-) with enough RAM we could probably serve 10K sessions, though that would require 10K ppp daemons until we got the kernel bypass working. in either case it would presently leave a HUGE ifconfig -a output as there would be 10000 intefeaces, each representing a channel to a client. It migh tbe worth wondering if we need to abstract a 'class of PVCs' type device that feeds to some othe rway of splitting them. (we can do this with netgraph pretty easily.) (is it worth doing?) Julian > > Dave. > > -- > ============================================================================ > |David Gilbert, Velocet Communications. | Two things can only be | > |Mail: dgilbert@velocet.net | equal if and only if they | > |http://www.velocet.net/~dgilbert | are precisely opposite. | > =========================================================GLO================ > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-hackers" in the body of the message -- +------------------------------------+ ______ _ __ | __--_|\ Julian Elischer | \ U \/ / On assignment | / \ julian@Whistle.com +------>x USA \ in a strange | ( OZ ) 110 Marsh dr. Foster City, CA \___ ___ | country ! +- X_.---._/ USA+(650) 577-7063 (wk) \_/ \\ v To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message