From owner-freebsd-questions Tue Jun 1 5:53:39 1999 Delivered-To: freebsd-questions@freebsd.org Received: from europe.std.com (europe.std.com [199.172.62.20]) by hub.freebsd.org (Postfix) with ESMTP id 5E1DC15060 for ; Tue, 1 Jun 1999 05:53:35 -0700 (PDT) (envelope-from lowell@world.std.com) Received: from world.std.com by europe.std.com (STD1.2/BZS-8-1.0) id IAA21919; Tue, 1 Jun 1999 08:53:29 -0400 (EDT) Received: by world.std.com (TheWorld/Spike-2.0) id AA08240; Tue, 1 Jun 1999 08:53:29 -0400 To: freebsd-questions@freebsd.org Subject: Re: FreeBSD as a Dedicated Router Organization: myself, not my employer -- which is *not* Software Tool & Die, anyway... References: From: Lowell Gilbert Date: 01 Jun 1999 08:53:29 -0400 In-Reply-To: Doug White's message of Mon, 31 May 1999 21:50:50 -0700 (PDT) Message-Id: Lines: 41 X-Mailer: Gnus v5.5/Emacs 20.2 Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Doug White writes: > On Fri, 28 May 1999, O'Connell wrote: > > > FAQ 160 alludes to deficiencies of FreeBSD as a dedicated LAN router in > > terms of good engineering practice and compliance with Internet standards. > > I'm not sure about the standards compliance bit, but the good engineering There are several optional features in RFC 1812 (router requirements) for which FreeBSD sets the defaults the wrong way. In all the cases I can think of, the RFC requirements are arguably outdated and FreeBSD's settings are reasonable. In a way, it's a shame that we can no longer assume that someone configuring a router understands IP. One example I can think of offhand is forwarding subnet broadcasts. If you disable this behaviour by default, you are technically noncompliant with 1812. Nonetheless, it's a bad idea these days. The warning in the FAQ answer is reasonable, but so were the decisions to which it refers. > bit is good -- the PC architecture doesn't have the bandwidth to handle > the kind of data routers normally see. Plus, you can't hot-swap > components. The amount of bandwidth that the NICs can handle is basically the performance limit for a FreeBSD-based router. Which is to say that FreeBSD is not, itself, a limit at all. Well, okay, on my 17 MHz 486, the CPU can't keep up with the NIC's theoretical limit, but that's a *really* slow machine these days. > I wouldn't suggest it for a core router, but for a small office router on > up it should be OK. Good summary of the performance issues. In my own opinion, I don't think anything that does its forwarding in software is fast enough for the Internet core. But then again, I work on stuff that *is* meant for the core. Be well. Lowell Gilbert To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message