From owner-freebsd-hackers Mon Mar 19 23: 4:53 2001 Delivered-To: freebsd-hackers@freebsd.org Received: from list.framfab.se (list.framfab.se [195.54.96.202]) by hub.freebsd.org (Postfix) with ESMTP id 7C39337B718 for ; Mon, 19 Mar 2001 23:04:48 -0800 (PST) (envelope-from Marten.Wikstrom@framfab.se) Received: from stoent001.framfab.se (mail.sto.framfab.se [172.16.200.241]) by list.framfab.se (8.9.3/8.9.3) with ESMTP id IAA06398; Tue, 20 Mar 2001 08:58:10 +0100 Received: by STOENT001 with Internet Mail Service (5.5.2653.19) id ; Tue, 20 Mar 2001 08:04:38 +0100 Message-ID: From: =?iso-8859-1?Q?M=E5rten_Wikstr=F6m?= To: "'Dennis'" Cc: "'freebsd-hackers@freebsd.org'" Subject: RE: Routing latency Date: Tue, 20 Mar 2001 08:04:27 +0100 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG [snip] > >triggers every second and steals too much cpu. So my=20 > question is, how can I > >decrease this routing delay? > Were you loading the interface, or just passing nominal=20 > streams? What pps=20 > did you pass through the box? Most likely the "delays" are=20 > only seen when=20 > the machine is close to capacity (the slow CPU you are using=20 > doesnt help). I sent 20000 packets/s, three UDP streams with 60, 200 and 1000 bytes = sized packets respectively. I also tried just one stream with 60 bytes = packets and the same behaviour occured. > Latency under load and general latency are very different. Differing=20 > methods of handling backup conditions may have different=20 > goals; the proper=20 > goal is overall stability and NOT packet efficiency. It=20 > doesnt matter how=20 > fast a man runs if he doesnt finish the race. =20 > The problem with LINUX is that it works to a point and then=20 > chokes, while=20 > freebsd works up to higher thresholds. You cant evaluate a=20 > subsystem with=20 > one somewhat bogus test, without looking at the system as a whole. Yep, that is exactly what my test showed when I tested the packet = throughput capacity. Linux choked at 27000 pps and then the output rate = _decreased_ with higher input rate, whereas the FreeBSD box started to drop packets = at 19000 packets/s but the throughput did still increase up untill approximately 40000 pps. (output rate). The input rate was then 70000 = pps. > If you are using the dc driver, make certain it is operating in=20 > store-and-forward mode, the default configuration starts in a=20 > mode that=20 > only works on 10mb/s connections. I'm using the de driver. Alas, the NICs seems quite old. They are = 21140's. I've only got one 21143. I think there is a 3COM 3c905b in the lab too. Would it be better to use the 21143 + 3com than two 21140s? /M=E5rten To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message