From owner-freebsd-current@FreeBSD.ORG Thu Dec 23 22:45:24 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 9EA8416A4CE for ; Thu, 23 Dec 2004 22:45:24 +0000 (GMT) Received: from mail6.speakeasy.net (mail6.speakeasy.net [216.254.0.206]) by mx1.FreeBSD.org (Postfix) with ESMTP id 69D0443D1F for ; Thu, 23 Dec 2004 22:45:24 +0000 (GMT) (envelope-from jmg@hydrogen.funkthat.com) Received: (qmail 1866 invoked from network); 23 Dec 2004 22:45:24 -0000 Received: from gate.funkthat.com (HELO hydrogen.funkthat.com) ([69.17.45.168]) (envelope-sender ) by mail6.speakeasy.net (qmail-ldap-1.03) with SMTP for ; 23 Dec 2004 22:45:24 -0000 Received: from hydrogen.funkthat.com (clejqd@localhost.funkthat.com [127.0.0.1])iBNMjMGH056395; Thu, 23 Dec 2004 14:45:22 -0800 (PST) (envelope-from jmg@hydrogen.funkthat.com) Received: (from jmg@localhost) by hydrogen.funkthat.com (8.12.10/8.12.10/Submit) id iBNMjLa5056394; Thu, 23 Dec 2004 14:45:21 -0800 (PST) Date: Thu, 23 Dec 2004 14:45:21 -0800 From: John-Mark Gurney To: Maxim Sobolev Message-ID: <20041223224521.GI19624@funkthat.com> Mail-Followup-To: Maxim Sobolev , Poul-Henning Kamp , "current@freebsd.org" References: <97664.1102797317@critter.freebsd.dk> <41BD850A.5060001@portaone.com> <41BDEBFB.7010204@portaone.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <41BDEBFB.7010204@portaone.com> User-Agent: Mutt/1.4.1i X-Operating-System: FreeBSD 4.2-RELEASE i386 X-PGP-Fingerprint: B7 EC EF F8 AE ED A7 31 96 7A 22 B3 D8 56 36 F4 X-Files: The truth is out there X-URL: http://resnet.uoregon.edu/~gurney_j/ X-Resume: http://resnet.uoregon.edu/~gurney_j/resume.html cc: Poul-Henning Kamp cc: "current@freebsd.org" Subject: Re: GBDE write performance really sucks X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: John-Mark Gurney List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 23 Dec 2004 22:45:24 -0000 Maxim Sobolev wrote this message on Mon, Dec 13, 2004 at 21:22 +0200: > Little more investigation revealed that the problem was due to disabled > write cache in ata(4). It would be interesting to compare FreeBSD > behaviour to behaviour of other operating systems in such situation, > since the drop of sequental writing performance in the case of 8KB > blocks and disabled write cache in FreeBSD is about 20x (from more than > 20MB/sec to merely 1MB/sec), which doesn't look reasonably to me. I've heard (though not confirmed) that many hard disks use larger than 512byte blocks for on disk data.. If this is the case, when you turn off write cacheing you maybe requiring the hard disk to do a read/write of the same block significantly impacting performance... Don't forget that doing purely sequental writes introduces a bandwidth product into the mix.. If it always takes your hd 5ms to respond to a command, and you only do 8KB/sec, then we end up with a max data rate of 1600KB. With write caching disabled, you prevent the hd from optimizing how it does the writes. -- John-Mark Gurney Voice: +1 415 225 5579 "All that I will do, has been done, All that I have, has not."