From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261762AbULHGud (ORCPT ); Wed, 8 Dec 2004 01:50:33 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261991AbULHGud (ORCPT ); Wed, 8 Dec 2004 01:50:33 -0500 Received: from ns.virtualhost.dk ([195.184.98.160]:54937 "EHLO virtualhost.dk") by vger.kernel.org with ESMTP id S261762AbULHGu1 (ORCPT ); Wed, 8 Dec 2004 01:50:27 -0500 Date: Wed, 8 Dec 2004 07:49:29 +0100 From: Jens Axboe To: Nick Piggin Cc: Andrea Arcangeli , Andrew Morton , linux-kernel@vger.kernel.org Subject: Re: Time sliced CFQ io scheduler Message-ID: <20041208064925.GD3035@suse.de> References: <20041202130457.GC10458@suse.de> <20041202134801.GE10458@suse.de> <20041202114836.6b2e8d3f.akpm@osdl.org> <20041202195232.GA26695@suse.de> <20041208003736.GD16322@dualathlon.random> <1102467253.8095.10.camel@npiggin-nld.site> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1102467253.8095.10.camel@npiggin-nld.site> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Dec 08 2004, Nick Piggin wrote: > On Wed, 2004-12-08 at 01:37 +0100, Andrea Arcangeli wrote: > > On Thu, Dec 02, 2004 at 08:52:36PM +0100, Jens Axboe wrote: > > > with its default io scheduler has basically zero write performance in > > > > IMHO the default io scheduler should be changed to cfq. as is all but > > general purpose so it's a mistake to leave it the default (plus as Jens > > I think it is actually pretty good at general purpose stuff. For > example, the old writes starve reads thing. It is especially bad > when doing small dependent reads like `find | xargs grep`. (Although > CFQ is probably better at this than deadline too). Time sliced cfq fixes this. > It also tends to degrade more gracefully under memory load because > it doesn't require much readahead. Ditto. > > found the write bandwidth is not existent during reads, no surprise it > > falls apart in any database load). We had to make the cfq the default > > for the enterprise release already. The first thing I do is to add > > elevator=cfq on a new install. I really like how well cfq has been > > designed, implemented and turned, Jens's results with his last patch are > > quite impressive. > > > > That is synch write bandwidth. Yes that seems to be a problem. A pretty big one :-) -- Jens Axboe