From owner-freebsd-isp@FreeBSD.ORG Sun Jan 15 01:14:13 2006 Return-Path: X-Original-To: freebsd-isp@FreeBSD.org Delivered-To: freebsd-isp@FreeBSD.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 0C37016A41F for ; Sun, 15 Jan 2006 01:14:13 +0000 (GMT) (envelope-from glebius@FreeBSD.org) Received: from cell.sick.ru (cell.sick.ru [217.72.144.68]) by mx1.FreeBSD.org (Postfix) with ESMTP id 8E66043D62 for ; Sun, 15 Jan 2006 01:14:08 +0000 (GMT) (envelope-from glebius@FreeBSD.org) Received: from cell.sick.ru (glebius@localhost [127.0.0.1]) by cell.sick.ru (8.13.3/8.13.3) with ESMTP id k0F1E65B014620 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Sun, 15 Jan 2006 04:14:06 +0300 (MSK) (envelope-from glebius@FreeBSD.org) Received: (from glebius@localhost) by cell.sick.ru (8.13.3/8.13.1/Submit) id k0F1E0I9014619; Sun, 15 Jan 2006 04:14:00 +0300 (MSK) (envelope-from glebius@FreeBSD.org) X-Authentication-Warning: cell.sick.ru: glebius set sender to glebius@FreeBSD.org using -f Date: Sun, 15 Jan 2006 04:14:00 +0300 From: Gleb Smirnoff To: Danial Thom Message-ID: <20060115011400.GM83922@FreeBSD.org> References: <20060111133229.GF98918@over-yonder.net> <20060111134814.19609.qmail@web33307.mail.mud.yahoo.com> Mime-Version: 1.0 Content-Type: text/plain; charset=koi8-r Content-Disposition: inline In-Reply-To: <20060111134814.19609.qmail@web33307.mail.mud.yahoo.com> User-Agent: Mutt/1.5.6i Cc: freebsd-isp@FreeBSD.org, ann kok , "Matthew D. Fuller" Subject: Re: freebsd router X-BeenThere: freebsd-isp@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Internet Services Providers List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 15 Jan 2006 01:14:13 -0000 On Wed, Jan 11, 2006 at 05:48:14AM -0800, Danial Thom wrote: D> I'd be interested in hearing your reasoning for D> thinking so. There is little argument that D> FreeBSD 4.x is perhaps the fastest Uniprocessor D> O/S ever created for networking. SMP will likely D> never be able to match it. It certainly can't D> now, in the current state of development. D> D> Routing is fastest when implemented as a single D> process task. Once you start chopping up D> (threading) the path you slow it down. While it D> could be possible to have a faster routing D> subsystem on a custom-designed MP O/S, its not D> practical to build a general purpose O/S in such D> a way. D> D> So freebsd 4.x it is. Freebsd 4.x can route 25% D> more traffic than its 5.x counterpart on the same D> hardware. 5.x SMP is actually worse (as it drops D> more packets at high traffic levels, and FreeBSD D> 4.x never drops packets until its overrun). Do you have more exact information? I mean: - Description of the test setup. - How packet stream was generated? - How success/loss was measured? - What hardware was used: CPU, mobo, NICs. - What settings were non-default. - And finally exact numbers - pps success/loss. And don't waste your time comparing 5.x and 4.x. Please compare 4.x and 6.0. The 5.x is a previous step. -- Totus tuus, Glebius. GLEBIUS-RIPN GLEB-RIPE