From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: Optimize RAID0 for max IOPS? Date: Tue, 25 Jan 2011 16:35:23 -0500 Message-ID: <20110125213523.GA14375@infradead.org> References: <20110118210112.D13A236C@gemini.denx.de> <20110125171017.GA24921@infradead.org> <20110125184115.1119FB187@gemini.denx.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20110125184115.1119FB187@gemini.denx.de> Sender: linux-raid-owner@vger.kernel.org To: Wolfgang Denk Cc: Christoph Hellwig , linux-raid@vger.kernel.org List-Id: linux-raid.ids On Tue, Jan 25, 2011 at 07:41:15PM +0100, Wolfgang Denk wrote: > > I don't think you even want a RAID0 in that case. For small IOPs > > you're much better off with a simple concatenation of devices. > > What exactly do you mean by "conatenation"? LVM striping? > At least the discussion here does not show any significant advantages > for this concept: > http://groups.google.com/group/ubuntu-user-community/web/pick-your-pleasure-raid-0-mdadm-striping-or-lvm-striping No, concatenation means not using any striping, but just concatenating the disk linearly, e.g. +-----------------------------------+ | Filesystem | +--------+--------+--------+--------+ | Disk 1 | Disk 2 | Disk 3 | Disk 4 | +--------+--------+--------+--------+ This can be done using the using the MD linear target, or simply by having multiple PVs in a VG with LVM. > > > Make sure to disable the disk write caches and if not using the newest > > kernel also mount the filesystem with -o nobarrier. With lots of small > > I/Os and metadata intensive workloads that's usually a lot faster. > > Tests if done recently indicate that on the other hand nobarrier causes > a serious degradation of read and write performance (down to some 40% > of the values before). Do you have a pointer to your results?