From: Jan Kara Subject: Re: [PATCH] improve the performance of large sequential write NFS workloads Date: Tue, 22 Dec 2009 13:55:32 +0100 Message-ID: <20091222125532.GC604@atrey.karlin.mff.cuni.cz> References: <1261015420.1947.54.camel@serenity> <1261037877.27920.36.camel@laptop> <20091219122033.GA11360@localhost> <1261232747.1947.194.camel@serenity> <20091222122557.GA604@atrey.karlin.mff.cuni.cz> <1261485515.4937.28.camel@laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Jan Kara , Steve Rago , Wu Fengguang , "linux-nfs@vger.kernel.org" , "linux-kernel@vger.kernel.org" , "Trond.Myklebust@netapp.com" , "jens.axboe" , Peter Staubach To: Peter Zijlstra Return-path: Received: from atrey.karlin.mff.cuni.cz ([195.113.26.193]:38844 "EHLO atrey.karlin.mff.cuni.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750711AbZLVMzd (ORCPT ); Tue, 22 Dec 2009 07:55:33 -0500 In-Reply-To: <1261485515.4937.28.camel@laptop> Sender: linux-nfs-owner@vger.kernel.org List-ID: > On Tue, 2009-12-22 at 13:25 +0100, Jan Kara wrote: > > I believe that if we had per-bdi dirty_background_ratio and set it low > > for NFS's bdi, > > There's two things there I think: > 1) bdi_background > 2) different background ratios per bdi Right. > 1) could be 'trivially' done much like we do bdi_dirty in > get_dirty_limits(). > > 2) I'm not at all convinced we want to go there. Yeah. Doing 1) and playing with bdi->min_ratio and bdi->max_ratio should be the first thing we should try... Honza -- Jan Kara SuSE CR Labs