From owner-freebsd-fs@FreeBSD.ORG Sun Mar 28 10:45:03 2010 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 09C3A106566B for ; Sun, 28 Mar 2010 10:45:03 +0000 (UTC) (envelope-from freebsd-fs@m.gmane.org) Received: from lo.gmane.org (lo.gmane.org [80.91.229.12]) by mx1.freebsd.org (Postfix) with ESMTP id B8EBC8FC0C for ; Sun, 28 Mar 2010 10:45:02 +0000 (UTC) Received: from list by lo.gmane.org with local (Exim 4.69) (envelope-from ) id 1NvpzH-0007eW-RM for freebsd-fs@freebsd.org; Sun, 28 Mar 2010 12:44:59 +0200 Received: from 93-138-119-5.adsl.net.t-com.hr ([93.138.119.5]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Sun, 28 Mar 2010 12:44:59 +0200 Received: from ivoras by 93-138-119-5.adsl.net.t-com.hr with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Sun, 28 Mar 2010 12:44:59 +0200 X-Injected-Via-Gmane: http://gmane.org/ To: freebsd-fs@freebsd.org From: Ivan Voras Date: Sun, 28 Mar 2010 12:44:35 +0200 Lines: 21 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@dough.gmane.org X-Gmane-NNTP-Posting-Host: 93-138-119-5.adsl.net.t-com.hr User-Agent: Thunderbird 2.0.0.21 (X11/20090612) In-Reply-To: Subject: Re: ZFS raidz and 4k sector disks X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 28 Mar 2010 10:45:03 -0000 Alexey Tarasov wrote: > Hello. > > I have reviewed a lot of discussion about new WD 4k sector disks (...EARS). > I have RAIDZ pool of such disks with very bad performance. Now my GPT ZFS partitions don't start from value dividable by 4 (162). > Some guys noticed that aligning ZFS partitions according to recommendation wouldn't help at all because RAIDZ uses variable stripe size. Yes, it doesn't group data in aligned "clusters" - the basic data alignment is sector-sized. > So where is the bottleneck of this configuration: 1) in ZFS which doesn't know about 4k sectorsize? 2) maybe somewhere inside FreeBSD VFS code or disk driver code? 3) somewhere else? AFAIK (I don't have actual experience with them) current 4k drives emulate 512b drives and have a performance penalty in the above scenario. Because of this emulation, ZFS doesn't know you have a 4k drive. If you can, try disabling this emulation and make it present to the operating system as a true 4k drive. Of course, this will make the data on it unreadable - you will have to reformat it, and the drive will be unbootable. Good luck and report what you find.