From: Anthony Liguori <anthony@codemonkey.ws>
To: Ryan Harper <ryanh@us.ibm.com>
Cc: Chris Wright <chrisw@redhat.com>,
Mark McLoughlin <markmc@redhat.com>,
kvm-devel <kvm-devel@lists.sourceforge.net>,
Laurent Vivier <Laurent.Vivier@bull.net>,
"qemu-devel@nongnu.org" <qemu-devel@nongnu.org>
Subject: [Qemu-devel] Re: [RFC] Disk integrity in QEMU
Date: Mon, 13 Oct 2008 13:43:14 -0500 [thread overview]
Message-ID: <48F396C2.30704@codemonkey.ws> (raw)
In-Reply-To: <20081013170610.GF21410@us.ibm.com>
Ryan Harper wrote:
> * Anthony Liguori <anthony@codemonkey.ws> [2008-10-09 12:00]:
>
>> Read performance should be unaffected by using O_DSYNC. O_DIRECT will
>> significantly reduce read performance. I think we should use O_DSYNC by
>> default and I have sent out a patch that contains that. We will follow
>> up with benchmarks to demonstrate this.
>>
>>
With 16k writes I think we hit a pathological case with the particular
storage backend we're using since it has many disks and the volume is
striped. Also the results a bit different when going through a file
system verses a LVM partition (the later being the first data set).
Presumably, this is because even with no flags, writes happen
synchronously to a LVM partition.
Also, cache=off seems to do pretty terribly when operating on an ext3
file. I suspect this has to do with how ext3 implements O_DIRECT.
However, the data demonstrates pretty nicely that O_DSYNC gives you
native write speed, but accelerated read speed which I think we agree is
the desirable behavior. cache=off never seems to outperform cache=wt
which is another good argument for it being the default over cache=off.
Regards,
Anthony Liguori
next prev parent reply other threads:[~2008-10-13 18:43 UTC|newest]
Thread overview: 101+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-10-09 17:00 [Qemu-devel] [RFC] Disk integrity in QEMU Anthony Liguori
2008-10-10 7:54 ` Gerd Hoffmann
2008-10-10 8:12 ` Mark McLoughlin
2008-10-12 23:10 ` Jamie Lokier
2008-10-14 17:15 ` Avi Kivity
2008-10-10 9:32 ` Avi Kivity
2008-10-12 23:00 ` Jamie Lokier
2008-10-10 8:11 ` Aurelien Jarno
2008-10-10 12:26 ` Anthony Liguori
2008-10-10 12:53 ` Paul Brook
2008-10-10 13:55 ` Anthony Liguori
2008-10-10 14:05 ` Paul Brook
2008-10-10 14:19 ` Avi Kivity
2008-10-17 13:14 ` Jens Axboe
2008-10-19 9:13 ` Avi Kivity
2008-10-10 15:48 ` Aurelien Jarno
2008-10-10 9:16 ` Avi Kivity
2008-10-10 9:58 ` Daniel P. Berrange
2008-10-10 10:26 ` Avi Kivity
2008-10-10 12:59 ` Paul Brook
2008-10-10 13:20 ` Avi Kivity
2008-10-10 12:34 ` Anthony Liguori
2008-10-10 12:56 ` Avi Kivity
2008-10-11 9:07 ` andrzej zaborowski
2008-10-11 17:54 ` Mark Wagner
2008-10-11 20:35 ` Anthony Liguori
2008-10-12 0:43 ` Mark Wagner
2008-10-12 1:50 ` Chris Wright
2008-10-12 16:22 ` Jamie Lokier
2008-10-12 17:54 ` Anthony Liguori
2008-10-12 18:14 ` nuitari-qemu
2008-10-13 0:27 ` Mark Wagner
2008-10-13 1:21 ` Anthony Liguori
2008-10-13 2:09 ` Mark Wagner
2008-10-13 3:16 ` Anthony Liguori
2008-10-13 6:42 ` Aurelien Jarno
2008-10-13 14:38 ` Steve Ofsthun
2008-10-12 0:44 ` Chris Wright
2008-10-12 10:21 ` Avi Kivity
2008-10-12 14:37 ` Dor Laor
2008-10-12 15:35 ` Jamie Lokier
2008-10-12 18:00 ` Anthony Liguori
2008-10-12 18:02 ` Anthony Liguori
2008-10-15 10:17 ` Andrea Arcangeli
2008-10-12 17:59 ` Anthony Liguori
2008-10-12 18:34 ` Avi Kivity
2008-10-12 19:33 ` Izik Eidus
2008-10-14 17:08 ` Avi Kivity
2008-10-12 19:59 ` Anthony Liguori
2008-10-12 20:43 ` Avi Kivity
2008-10-12 21:11 ` Anthony Liguori
2008-10-14 15:21 ` Avi Kivity
2008-10-14 15:32 ` Anthony Liguori
2008-10-14 15:43 ` Avi Kivity
2008-10-14 19:25 ` Laurent Vivier
2008-10-16 9:47 ` Avi Kivity
2008-10-12 10:12 ` Avi Kivity
2008-10-17 13:20 ` Jens Axboe
2008-10-19 9:01 ` Avi Kivity
2008-10-19 18:10 ` Jens Axboe
2008-10-19 18:23 ` Avi Kivity
2008-10-19 19:17 ` M. Warner Losh
2008-10-19 19:31 ` Avi Kivity
2008-10-19 18:24 ` Avi Kivity
2008-10-19 18:36 ` Jens Axboe
2008-10-19 19:11 ` Avi Kivity
2008-10-19 19:30 ` Jens Axboe
2008-10-19 20:16 ` Avi Kivity
2008-10-20 14:14 ` Avi Kivity
2008-10-10 10:03 ` Fabrice Bellard
2008-10-13 16:11 ` Laurent Vivier
2008-10-13 16:58 ` Anthony Liguori
2008-10-13 17:36 ` Jamie Lokier
2008-10-13 17:06 ` [Qemu-devel] " Ryan Harper
2008-10-13 18:43 ` Anthony Liguori [this message]
2008-10-14 16:42 ` Avi Kivity
2008-10-13 18:51 ` Laurent Vivier
2008-10-13 19:43 ` Ryan Harper
2008-10-13 20:21 ` Laurent Vivier
2008-10-13 21:05 ` Ryan Harper
2008-10-15 13:10 ` Laurent Vivier
2008-10-16 10:24 ` Laurent Vivier
2008-10-16 13:43 ` Anthony Liguori
2008-10-16 16:08 ` Laurent Vivier
2008-10-17 12:48 ` Avi Kivity
2008-10-17 13:17 ` Laurent Vivier
2008-10-14 10:05 ` Kevin Wolf
2008-10-14 14:32 ` Ryan Harper
2008-10-14 16:37 ` Avi Kivity
2008-10-13 19:00 ` Mark Wagner
2008-10-13 19:15 ` Ryan Harper
2008-10-14 16:49 ` Avi Kivity
2008-10-13 17:58 ` [Qemu-devel] " Rik van Riel
2008-10-13 18:22 ` Jamie Lokier
2008-10-13 18:34 ` Rik van Riel
2008-10-14 1:56 ` Jamie Lokier
2008-10-14 2:28 ` nuitari-qemu
2008-10-28 17:34 ` Ian Jackson
2008-10-28 17:45 ` Anthony Liguori
2008-10-28 17:50 ` Ian Jackson
2008-10-28 18:19 ` Jamie Lokier
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=48F396C2.30704@codemonkey.ws \
--to=anthony@codemonkey.ws \
--cc=Laurent.Vivier@bull.net \
--cc=chrisw@redhat.com \
--cc=kvm-devel@lists.sourceforge.net \
--cc=markmc@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=ryanh@us.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).