From owner-freebsd-stable@FreeBSD.ORG Wed Jul 25 20:51:08 2007 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 1F73D16A418 for ; Wed, 25 Jul 2007 20:51:08 +0000 (UTC) (envelope-from scottl@samsco.org) Received: from pooker.samsco.org (pooker.samsco.org [168.103.85.57]) by mx1.freebsd.org (Postfix) with ESMTP id AC88E13C457 for ; Wed, 25 Jul 2007 20:51:07 +0000 (UTC) (envelope-from scottl@samsco.org) Received: from phobos.samsco.home (phobos.samsco.home [192.168.254.11]) (authenticated bits=0) by pooker.samsco.org (8.13.8/8.13.8) with ESMTP id l6PKp1PU014486; Wed, 25 Jul 2007 14:51:01 -0600 (MDT) (envelope-from scottl@samsco.org) Message-ID: <46A7B7AF.6080308@samsco.org> Date: Wed, 25 Jul 2007 14:50:55 -0600 From: Scott Long User-Agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en-US; rv:1.8.1.4) Gecko/20070509 SeaMonkey/1.1.2 MIME-Version: 1.0 To: Howard Goldstein References: <46A4E8FA.6010403@queue.to> <46A7B3FB.7010504@queue.to> In-Reply-To: <46A7B3FB.7010504@queue.to> X-Enigmail-Version: 0.95.1 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Greylist: Sender succeeded SMTP AUTH authentication, not delayed by milter-greylist-2.0.2 (pooker.samsco.org [168.103.85.57]); Wed, 25 Jul 2007 14:51:01 -0600 (MDT) X-Spam-Status: No, score=-1.4 required=5.5 tests=ALL_TRUSTED autolearn=failed version=3.1.8 X-Spam-Checker-Version: SpamAssassin 3.1.8 (2007-02-13) on pooker.samsco.org Cc: freebsd-stable@freebsd.org Subject: Re: [resolved, naively] Re: geom vs ich through ar device - benchmarks? X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 25 Jul 2007 20:51:08 -0000 Howard Goldstein wrote: > Howard Goldstein wrote: >> Has anyone done any benchmarks in desktop or server environment >> comparing geom with an ICH controller through the ar device in RAID1 >> service? Teh google, it seems to pick up grammar school math >> assignments lots of what may be relevant hits for fortunate speakers of >> German :( > > In case it may help someone else I've some benchmarks to share. These > are trivial and naive benchmarks since they test only one case which > probably is the worst case for geom. > > Testbed: Pair of WDC3200AAKS 320gb SATA, freshly newfsd 10gb filesystem > mounted with softupdates, remounted after each test > P4 @ 3ghz on a P4P800 in 6.2-STABLE, single user mode, ICH5R controller > detects these SATA-II drives inexplicably as SATA-I > ICH5 only support SATA-1. > Naive test: dd if=/dev/zero of=/fsundertest/tstfile bs=1m count=1000 > *not including buffered stuff unwritten, the test simply reported dd's > idea of the transfer time*, and then the other way where it doesn't matter > > In firmware's RAID1 Using the awesome gmirror > write 1gb: 13.275 13.7 > rd 12.9 13.8 > > Of course after this I used gmirror... Just so we're clear, the ICH5 doesn't have any firmware and doesn't actually do any RAID operations. What is has is hook into the system BIOS during boot. That hook allows the BIOS to do RAID-like operations during boot, until the OS takes over control of the devices. After that, it's up to the OS to do all the RAID work. The 'ar' driver is still software RAID, just like gmirror. What you've effectively done merely compare the performance of one software RAID stack to another. That's certainly an interesting comparison, but maybe not exactly what you had in mind. Scott