From: Jens Axboe Subject: Re: [patch/rft] jbd2: tag journal writes as metadata I/O Date: Fri, 2 Apr 2010 09:00:02 +0200 Message-ID: <20100402070001.GY23510@kernel.dk> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org, esandeen@redhat.com To: Jeff Moyer Return-path: Received: from 0122700014.0.fullrate.dk ([95.166.99.235]:51380 "EHLO kernel.dk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759213Ab0DBHAD (ORCPT ); Fri, 2 Apr 2010 03:00:03 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-ext4-owner@vger.kernel.org List-ID: On Thu, Apr 01 2010, Jeff Moyer wrote: > Hi, > > In running iozone for writes to small files, we noticed a pretty big > discrepency between the performance of the deadline and cfq I/O > schedulers. Investigation showed that I/O was being issued from 2 > different contexts: the iozone process itself, and the jbd2/sdh-8 thread > (as expected). Because of the way cfq performs slice idling, the delays > introduced between the metadata and data I/Os were significant. For > example, cfq would see about 7MB/s versus deadline's 35 for the same > workload. I also tested fs_mark with writing and fsyncing 1000 64k > files, and a similar 5x performance difference was observed. Eric > Sandeen suggested that I flag the journal writes as metadata, and once I > did that, the performance difference went away completely (cfq has > special logic to prioritize metadata I/O). > > So, I'm submitting this patch for comments and testing. I have a > similar patch for jbd that I will submit if folks agree that this is a > good idea. Looks good to me. -- Jens Axboe