From owner-freebsd-stable@FreeBSD.ORG Fri Jul 27 17:34:34 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 3D8CB16A41A for ; Fri, 27 Jul 2007 17:34:34 +0000 (UTC) (envelope-from hg@queue.to) Received: from pickle.queue.to (pickle.queue.to [71.180.69.18]) by mx1.freebsd.org (Postfix) with ESMTP id C005813C45A for ; Fri, 27 Jul 2007 17:34:33 +0000 (UTC) (envelope-from hg@queue.to) Received: (qmail 12245 invoked from network); 27 Jul 2007 13:34:32 -0400 Received: from cally.queue.to (172.16.0.6) by pickle.queue.to with ESMTP; 27 Jul 2007 13:34:32 -0400 Message-ID: <46AA2CA8.2080009@queue.to> Date: Fri, 27 Jul 2007 13:34:32 -0400 From: Howard Goldstein User-Agent: Thunderbird 2.0.0.5 (X11/20070720) MIME-Version: 1.0 To: freebsd-stable@freebsd.org References: <46A4E8FA.6010403@queue.to> <46A7B3FB.7010504@queue.to> In-Reply-To: <46A7B3FB.7010504@queue.to> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit 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: Fri, 27 Jul 2007 17:34:34 -0000 Howard Goldstein wrote: > > ar device's RAID1 Using the awesome gmirror > write 1gb: 13.275 13.7 > rd 12.9 13.8 I recently had an opportunity to repeat this test with the same exact WD3200AAKS (7200RPM consumer grade 320g sata II) drives on an older hardware RAID card, the 3ware 8006-2LP plugged into a 32 bit PCI slot. The results were surprisingly poor given that this card does the RAID in hardware. With RAID-1 Write 1gb: 24.3s Read 1gb: 18.6 Using the same dd invocation as in the previous message, to a freshly newfsd 10GB slice at the beginning of the drive. The drives were strapped to 150MB/S since the 8006 has issues with SATA-II and the interface is nowhere near saturating anyway since the ultimate limit of the head<>platter is reached well below 150MB/S Sticking with gmirror/geom (thank you Pawel) FYI...