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] Oracle on a classic Physical server Win2k3 versusOracle

To: Rémi BERNIER <R.BERNIER@xxxxxxxxxxx>
Subject: Re: [Xen-users] Oracle on a classic Physical server Win2k3 versusOracle on domU win2k3
From: Pasi Kärkkäinen <pasik@xxxxxx>
Date: Thu, 5 Nov 2009 11:48:31 +0200
Cc: James Harper <james.harper@xxxxxxxxxxxxxxxx>, Xen-users@xxxxxxxxxxxxxxxxxxx
Delivery-date: Thu, 05 Nov 2009 01:49:22 -0800
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <OF28189488.13645B8E-ONC1257665.0029B793-C1257665.002B0486@xxxxxxxxxxx>
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: <OF97D32B53.7F27D4CB-ONC1257662.00383EC4-C1257662.0039709B@xxxxxxxxxxx> <OF28189488.13645B8E-ONC1257665.0029B793-C1257665.002B0486@xxxxxxxxxxx>
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Mutt/1.5.13 (2006-08-11)
On Thu, Nov 05, 2009 at 08:49:50AM +0100, Rémi BERNIER wrote:
>    Hi everyone,
> 
>    Bad news...
>    I just test with Xen 3.4.1 modules 2.6.26-6, but the performance are far
>    from being the same quality with a physical server win2k3
>    If anyone have another idea..
> 

You really need to profile _why_ it's slower. you need to measure
exactly _what_ is using the cpu, or causing the slowness.

You need to monitor in dom0 (using "xm top", "iostat 1", etc) and *also*
in the windows guest, using whatever tools windows has.

Also, why aren't you running oracle on Linux? would be a lot easier to
debug..

-- Pasi

>    Rémi
> 
>    Pasi,
>    I just finished the swingbench test with a dedicated CPU but it's the same
>    performance.
>    When i will have time, I will test on Xen 3.4.1
> 
>    On Mon, Nov 02, 2009 at 09:32:59AM +0100, Rémi BERNIER wrote:
>    >    Yes my windows guest is fully CPU but i thought  to have better
>    >    performance regarding the number of transactions per seconds compared
>    a
>    >    Physical server.
>    >    Thanks,
>    >
> 
>    Have you tried pinning the guest vcpus to separate physical cores? And
>    dedicating a separate core for dom0?
> 
>    Also, did you try with Xen 3.4.1 ?
> 
>    -- Pasi
> 
>    >    Rémi
>    >
>    >    On Mon, Nov 02, 2009 at 08:55:40AM +0100, Rémi BERNIER wrote:
>    >    >    James, Pasi,
>    >    >
>    >    >    Yes it's just a swinbench on the performance of transactions on
>    a
>    >    Oracle
>    >    >    database.
>    >    >    But not a network test.
>    >    >
>    >    >    My CPU is Intel(R) Xeon(R) CPU 5130 @ 2.00GHz, when i test with
>    >    >    swingbench, my domU is full cpu (near 197%)
>    >    >    The result of iowait with command: iostat 1  -> %iowait = 0.00
>    >    >
>    >    >    Thanks in advance.
>    >    >
>    >
>    >    Ok so it's fully CPU bound.
>    >
>    >    Then you should measure CPU usage in the Windows guest and figure out
>    >    where all the CPU time goes..
>    >
>    >    -- Pasi
>    >
>    >    >    Rémi
>    >    >
>    >    >    While running the test, run "xm top" in dom0.
>    >    >
>    >    >    How's the cpu usage? are your cpu cores maxed out?
>    >    >    What's using cpu?
>    >    >
>    >    >    Also run "iostat 1" in dom0 (it's in sysstat package), and see
>    if you
>    >    >    have a lot of iowait in dom0..
>    >    >
>    >    >    -- Pasi
>    >    >
>    >    >    >
>    >    >    > I just finished the swingbench test with new drivers
>    (0.10.0.130)
>    >    but
>    >    >    it's the
>    >    >    > same result about 8400 transactions per seconds..
>    >    >    > So if you have an idea, i'm listening to you.
>    >    >    >
>    >    >    > Thanks in advance and good week.
>    >    >    >
>    >    >
>    >    >    Is the swingbench test disk i/o bound? Or network too?
>    >    >
>    >    >    James
>    >    >
>    >    >    _______________________________________________
>    >    >    Xen-users mailing list
>    >    >    Xen-users@xxxxxxxxxxxxxxxxxxx
>    >    >    http://lists.xensource.com/xen-users
>    >
>    >    _______________________________________________
>    >    Xen-users mailing list
>    >    Xen-users@xxxxxxxxxxxxxxxxxxx
>    >    http://lists.xensource.com/xen-users
>    _______________________________________________
>    Xen-users mailing list
>    Xen-users@xxxxxxxxxxxxxxxxxxx
>    http://lists.xensource.com/xen-users

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