From owner-freebsd-questions Tue Jul 24 14:59:35 2001 Delivered-To: freebsd-questions@freebsd.org Received: from shell.bchosting.com (shell.bchosting.com [64.69.68.35]) by hub.freebsd.org (Postfix) with ESMTP id 9926537B403 for ; Tue, 24 Jul 2001 14:59:31 -0700 (PDT) (envelope-from chris@selkie.org) Received: from localhost (chris@localhost) by shell.bchosting.com (8.11.0/8.11.0) with ESMTP id f6OLxHR19699; Tue, 24 Jul 2001 14:59:17 -0700 (PDT) (envelope-from chris@selkie.org) Date: Tue, 24 Jul 2001 14:59:16 -0700 (PDT) From: Chris Phillips X-Sender: chris@shell.bchosting.com To: Chris Dillon Cc: Joe Clarke , freebsd-questions@FreeBSD.ORG Subject: Re: FreeBSD Router In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Hey Chris, I've decided to replace both the primary and secondary routers with PIII 1000s with 512MB of ram. I hope this helps. Thanks. -Chris Phillips On Tue, 24 Jul 2001, Chris Dillon wrote: > On Tue, 24 Jul 2001, Chris Phillips wrote: > > > As mentioned, I use Intel EtherExpress Pro 10/100's exclusively on > > all network servers. I don't imagine the NICs are a problem. > > > > Ideas? > > > The P166 you're using might be just a bit underpowered for three > 100Mbit links and 250 ipfw rules. > > I'm running 8 100Mbit full-duplex links (using fxp cards as well) with > 66 static ipfw rules with an average additional 300 dynamic rules on a > PIII-800 Xeon. Previously the exact same setup ran just fine on a > PIII-500. This box has quite a few cycles to spare most of the time. > Just now, while doing an FTP of a large file from one box to another > on different networks, top reports 25% of the CPU used on the router > (about 20% for interrupts). I averaged 11.22MBytes/sec for the file > transfer, which would equate to 90Mbit/sec just for the raw file, not > including any other overhead (TCP/IP, the Ethernet frames, etc.). > Just from a CPU usage perspective, assuming no other bottlenecks pop > up, I should be able to move about 400Mbit/sec between networks > (thinking in terms of switch backplane bandwidth... 100Mbits IN one > port and OUT another is 100Mbits, not 200Mbits). > > So, to be able to move at least 100Mbit/sec across your router (in > one, out another) with about 250 ipfw rules, I'd guesstimate you would > need at least a Pentium 233 MMX. Not that the MMX instructions help > any, just that the MMX Pentiums (P55C?) have a larger L1 cache than > previous Pentiums. A non-MMX Pentium probably isn't available at > 233MHz anyway. :-) > > > -- Chris Dillon - cdillon@wolves.k12.mo.us - cdillon@inter-linc.net > FreeBSD: The fastest and most stable server OS on the planet > - Available for IA32 (Intel x86) and Alpha architectures > - IA64 (Itanium), PowerPC, and ARM architectures under development > - http://www.freebsd.org > > > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message