WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-users

Re: [Xen-users] slow dom0 write operations with 2.6.32 pv-ops and Xen4.0

To: benco@xxxxxxx
Subject: Re: [Xen-users] slow dom0 write operations with 2.6.32 pv-ops and Xen4.0.0-rc4
From: Pasi Kärkkäinen <pasik@xxxxxx>
Date: Mon, 1 Mar 2010 13:43:43 +0200
Cc: xen-users@xxxxxxxxxxxxxxxxxxx
Delivery-date: Mon, 01 Mar 2010 03:44:58 -0800
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <20100301120709.GQ29834@xxxxxxx>
List-help: <mailto:xen-users-request@lists.xensource.com?subject=help>
List-id: Xen user discussion <xen-users.lists.xensource.com>
List-post: <mailto:xen-users@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=unsubscribe>
References: <20100228230420.GP29834@xxxxxxx> <20100301074037.GJ2761@xxxxxxxxxxx> <20100301120709.GQ29834@xxxxxxx>
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Mutt/1.5.18 (2008-05-17)
On Mon, Mar 01, 2010 at 01:07:10PM +0100, benco@xxxxxxx wrote:
> > You're measuring buffered (cached) performance, so when you limit
> > the amount of cache to 2 GB the performance will be slower.
> > 
> > Try adding "oflag=direct" do your dd commands and test again.
> 
> 
> oflag=direct option produce almost the same results on xen and non-xen
> systems:
> 
> #dd if=/dev/zero of=zerofile.tst bs=1k count=1000000 oflag=direct
> 1000000+0 records in
> 1000000+0 records out
> 1024000000 bytes (1.0 GB) copied, 616.219 s, 1.7 MB/s
> 

Try bigger block size and smaller count, for example bs=64k. 
1k block are really small and won't give you good performance.

> what is interesting for me is the fact that with no cpu/memory limit and with
> caching the result is approx. 60% slowdown with xen kernel in comparison
> with non-xen kernel. Is it normal?
> 

Yeah, it depends on a lot of things. Try with DIRECT io (and big enough block 
size), 
to get some real/meaningful numbers.

-- Pasi


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users