From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-out1.uio.no ([129.240.10.57]:53972 "EHLO mail-out1.uio.no" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760529Ab2EPTg2 (ORCPT ); Wed, 16 May 2012 15:36:28 -0400 Received: from mail-mx3.uio.no ([129.240.10.44]) by mail-out1.uio.no with esmtp (Exim 4.75) (envelope-from ) id 1SUjmf-00042o-J3 for fio@vger.kernel.org; Wed, 16 May 2012 21:21:17 +0200 Received: from 7.123.45.31.customer.cdi.no ([31.45.123.7] helo=heritage.uio.no) by mail-mx3.uio.no with esmtpsa (TLSv1:AES256-SHA:256) user jb (Exim 4.76) (envelope-from ) id 1SUjme-0006uc-UE for fio@vger.kernel.org; Wed, 16 May 2012 21:21:17 +0200 Message-ID: <4FB3FE27.3030302@usit.uio.no> Date: Wed, 16 May 2012 21:21:11 +0200 From: =?ISO-8859-1?Q?Jarle_Bj=F8rgeengen?= MIME-Version: 1.0 Subject: blktrace with fio replay for benchmarking vendor offerings Content-Type: text/plain; charset="iso-8859-1"; format="flowed" Content-Transfer-Encoding: quoted-printable Sender: fio-owner@vger.kernel.org List-Id: fio@vger.kernel.org To: fio@vger.kernel.org Hello, I'm involved in the purchasing process of block storage systems, and=20 research viable benchmarking strategies for specifying and verifying=20 performance requirements. Ideally I would like to capture our traces of our current daily=20 production workload with blktrace, attach the traces to the tender and=20 require the configuration to be able to run 2x that kind of workload.=20 During acceptance I would like to hook enough hardware to saturate the=20 system with the same workload, and measure that the requirement has been=20 met. I'm interested in comments about the practical viability of such=20 approach if anyone have similar experiences. Some open questions: How safe is it to run blktrace on critical production environment? What=20 precautions should be made? Given that the current system consists of 3 HP EVA800 with X number of=20 LUNS about 50 hosts, and the new system likely is a single system with 5=20 new servers running at at full speed, how much value will the benchmark=20 provide compared to "real world" ? Should I create equally many luns and=20 distribute load across the 5 machines? Is it best to scale the workload to 2X replaying all traces=20 simultanously with one fio-trace-replay/lun, and then dublicate alle=20 replays so that each lun serves two fio-replays rather than one, and=20 then 3 and so on. Any thoughts or comments are very much appreciated. Best regards Jarle Bj=EF=BF=BDrgeengen University of Oslo / USIT