From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nikola Ciprich Subject: Re: kvm-83 write performance raw Date: Tue, 3 Mar 2009 16:13:25 +0100 Message-ID: <20090303151324.GA7090@develbox.linuxbox.cz> References: <20090302205330.GC20969@netvalue.net.nz> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Mark van Walraven , kvm@vger.kernel.org, nikola.ciprich@linuxbox.cz To: Malinka Rellikwodahs Return-path: Received: from gw3.lbox.cz ([62.245.111.133]:60229 "EHLO develbox.linuxbox.cz" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751993AbZCCPmT (ORCPT ); Tue, 3 Mar 2009 10:42:19 -0500 Content-Disposition: inline In-Reply-To: Sender: kvm-owner@vger.kernel.org List-ID: Hi, I think DRBD *MIGHT* be Your problem anyways... Can You try repeating Your measurments with no-disk-barrier, no-disk-flushes, no-disk-drain options for Your drbd devices and report the results? nik On Mon, Mar 02, 2009 at 04:00:57PM -0500, Malinka Rellikwodahs wrote: > On Mon, Mar 2, 2009 at 15:53, Mark van Walraven wrote: > > On Mon, Mar 02, 2009 at 03:11:59PM -0500, Malinka Rellikwodahs wrot= e: > >> when running with a raw disk image as a file or a raw disk image o= n an > >> lvm vg, I'm getting very low performance on write (5-10 MB/s) howe= ver > >> when using qcow2 format disk image the write speed is much better > >> (~30MB/s), which is consistant with a very similar setup running > >> kvm-68. =A0Unfortunately when running the test with qcow2 the syst= em > >> becomes unresponsive for a brief time during the test. > > > >> The host is running raid5 and drbd (drive replication software), > >> however performance on the host is performaning well and avoiding = the > >> drbd layer in the guest does not improve performance, but running = on > >> qcow2 does. > >> > >> Any thoughts/suggestions of what could be wrong or what to do to f= ix this? > > > > RAID1 has *much* better write performance. =A0With striping RAIDs, = alignment > > is important. =A0RAID controllers sometimes introduce hidden alignm= ent > > offsets. =A0Excessive read-ahead is a waste of time with a lot of s= mall > > random I/O, which is what I see mostly with guests on flat disk ima= ges. > > > > With LVM, it pays to make sure the LVs are aligned to the disk. =A0= I prefer > > boundaries with multiples of at least 64-sectors, which makes the L= VM > > overhead virtually disappear. =A0I align the guest filesystems too,= when > > I can. > > > > I don't think DRBD has an effect on alignment, but you might look a= t > > keeping the metadata on another drive. > > > > Block - rather than file - images are much faster. > > > > Hope this helps, >=20 > It does, however unless I'm missing something the performance is bein= g > lost not in the lvm/raid/drbd config, because I'm using the same setu= p > for other partitions which are used for data on the host and write > performance to those drives is just fine. > -- > To unsubscribe from this list: send the line "unsubscribe kvm" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html >=20 --=20 ------------------------------------- Nikola CIPRICH LinuxBox.cz, s.r.o. 28. rijna 168, 709 01 Ostrava tel.: +420 596 603 142 fax: +420 596 621 273 mobil: +420 777 093 799 www.linuxbox.cz mobil servis: +420 737 238 656 email servis: servis@linuxbox.cz -------------------------------------