From owner-freebsd-net Fri Oct 18 14:41: 8 2002 Delivered-To: freebsd-net@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 082B937B401 for ; Fri, 18 Oct 2002 14:41:07 -0700 (PDT) Received: from mail.sandvine.com (sandvine.com [199.243.201.138]) by mx1.FreeBSD.org (Postfix) with ESMTP id 7511D43EA9 for ; Fri, 18 Oct 2002 14:41:06 -0700 (PDT) (envelope-from don@sandvine.com) Received: by mail.sandvine.com with Internet Mail Service (5.5.2653.19) id <42S9VA79>; Fri, 18 Oct 2002 17:41:05 -0400 Message-ID: From: Don Bowman To: "'Kevin_Stevens@pursued-with.net'" , Don Bowman Cc: freebsd-net@FreeBSD.ORG Subject: RE: ENOBUFS Date: Fri, 18 Oct 2002 17:40:57 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" 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 From: Kevin Stevens [mailto:Kevin_Stevens@pursued-with.net] > On Fri, 18 Oct 2002, Don Bowman wrote: > > > > what do you mean ? it works great for me. even on -current i > > > can push out over 400kpps (64byte frames) on a 2.4GHz box. > > > > 400kpps seems like very poor performance. > > Unless I do the math wrong, this is only ~200Mbps, > > the nic should be able to allow ~2-3Mpps (GE bidirectional). > > First, you're only pushing packets, so you are only talking a > potential 1GB, not two. > > Second, sending minimum-size packets, while a best-case > metric for pps, is a worst-case metric for throughput. > I don't think that you can conclude that 20% theoretical > bandwidth utilization at minimum packet size is poor > performance; in fact it seems pretty good to me. Problem is, I'm making a bridge application which operates in a network similar to a router, not an endpoint server which receives TCP packets. There's a huge number of small packets on the internet due to TCP ACK's etc. I need to be able to do ~1.5Mpps x 2 for a single interface (both directions), which becomes ~1.5Mpps x 4 for an 'in' and an 'out' interface. Clearly you are correct that for most client/server applications 64-byte pps are not the interesting stat. Clearly I will have some tuning ahead, and likely I will not succeed, but for sure my 1U XEON with 6 gigabit nics will work very hard for its living :) --don (don@sandvine.com www.sandvine.com) To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message