From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtprelay0087.hostedemail.com ([216.40.44.87]:51622 "EHLO smtprelay.hostedemail.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1750743AbaATV3w (ORCPT ); Mon, 20 Jan 2014 16:29:52 -0500 Message-ID: <52DD954E.7020706@nellans.org> Date: Mon, 20 Jan 2014 15:29:50 -0600 From: David Nellans MIME-Version: 1.0 Subject: Re: Strange aggregate BW References: <748581DD-9A26-41D7-A537-58F6D6741B33@gmail.com> In-Reply-To: <748581DD-9A26-41D7-A537-58F6D6741B33@gmail.com> Content-Type: text/plain; charset="windows-1252"; format="flowed" Content-Transfer-Encoding: quoted-printable Sender: fio-owner@vger.kernel.org List-Id: fio@vger.kernel.org To: grant <626542@gmail.com>, fio@vger.kernel.org On 01/20/2014 03:23 PM, grant wrote: > Hello, > > I=EF=BF=BDve got a HWRAID with 10 NL-SAS HDDs. > All of them used in logical volume RAID6. > I=EF=BF=BDve tried to benchmark created volume with FIO v2.1.4: > > # cat seqwrite.fio > [write] > blocksize=3D512k > filename=3D/dev/sda > rw=3Dwrite > direct=3D1 > buffered=3D0 > ioengine=3Dlibaio > iodepth=3D1 > runtime=3D10m > numjobs=3D16 > > #./fio seqwrite.fio > > FIO returns: > ... > Run status group 0 (all jobs): > WRITE: io=3D127958MB, aggrb=3D2132.5MB/s, minb=3D56530KB/s, maxb=3D229= 141KB/s.. Between some hardware raid buffering, large block transfers, and 100%=20 sequential write you'll certainly be near optimal for sequential write=20 I/O seen at each disk. 200MB/s/disk is higher than most folks would=20 probably ballpark, but its not off by an order of magnitude or anything.=20 I assume ~150MB/s sequential performance for most "good" consumer=20 7200RPM drives. Higher RPM alone could push you to ~200MB/s. Try turning it into random I/O and if it goes to shit, then the results are probably as-expected ;)