From owner-freebsd-current@FreeBSD.ORG Sat Jun 9 01:39:47 2007 Return-Path: X-Original-To: current@FreeBSD.org Delivered-To: freebsd-current@FreeBSD.ORG Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 57A2016A468; Sat, 9 Jun 2007 01:39:47 +0000 (UTC) (envelope-from bakul@bitblocks.com) Received: from mail.bitblocks.com (bitblocks.com [64.142.15.60]) by mx1.freebsd.org (Postfix) with ESMTP id 39BEC13C448; Sat, 9 Jun 2007 01:39:47 +0000 (UTC) (envelope-from bakul@bitblocks.com) Received: from bitblocks.com (localhost.bitblocks.com [127.0.0.1]) by mail.bitblocks.com (Postfix) with ESMTP id B97CD5B52; Fri, 8 Jun 2007 18:17:24 -0700 (PDT) To: Josef Karthauser In-reply-to: Your message of "Fri, 08 Jun 2007 23:35:00 BST." <20070608223500.GA1250@genius.tao.org.uk> Date: Fri, 08 Jun 2007 18:17:24 -0700 From: Bakul Shah Message-Id: <20070609011724.B97CD5B52@mail.bitblocks.com> Cc: current@FreeBSD.org Subject: Re: zfs drive configurations? X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 09 Jun 2007 01:39:47 -0000 > I've got a machine I'm putting together with 5 drive bays, and I'm > trying to work out what combination of drives to use. The appeal of zfs > is that I can start with relatively small drives, and replace them, on a > one-by-one basis, as they either die, or I get my hands on larger ones. > Is this possible? Yes. [In theory. I have not tried this yet] > What if I have 5 200gb drives configured as a raidz pool, and then I > replace one of the 200gb drives with a 400gb one. Operationally what > would I do? I believe you can do something like zpool replace The new device will get "resilvered" -- get all the missing data put on it and then brought up for operation. This can take a while. Note that you will not be able to use the extra disk until *all* the disks in a group (mirror or raid) have been replaced with bigger disks. Also realize that while raidz gives you more space, in general it will be slower than mirroring for random IO. If each disk does N random ops/sec, then a 4 disk raidz will given N random reads/sec, while a 2 stripe 2 way mirror will give you upto 4N random reads/sec. See http://blogs.sun.com/roch/category/ZFS http://www.solarisinternals.com/wiki/index.php/ZFS_Best_Practices_Guide BTW, you can experiment using memory devices backed with real files! With 5 disks I'd use four for one zfs pool and the last one for temp space, and other misc. things.