RE: [Xen-users] MSSQL and Xen performance

[Nathan Eisenberg <nathan@xxxxxxxxxxxxxxxx>]

Damn you, Centos, for your old repository versions.  I'd highly recommend 
looking at a newer version of XEN than 3.1 - there have been some major 
performance gains since then.

Best Regards
Nathan Eisenberg
Sr. Systems Administrator
Atlas Networks, LLC
support@xxxxxxxxxxxxxxxx
http://support.atlasnetworks.us/portal


-----Original Message-----
From: xen-users-bounces@xxxxxxxxxxxxxxxxxxx 
[mailto:xen-users-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of blackc2004
Sent: Friday, May 08, 2009 8:47 AM
To: xen-users@xxxxxxxxxxxxxxxxxxx
Subject: RE: [Xen-users] MSSQL and Xen performance


Thanks everyone for the tips!

Unfortunately I didn't get to "pick" the consultants, they came with the
company that is doing our new ERP installation. So far I have not been happy
with the way things are going.

I am running:
Xen 3.1.2-128.1.6.el5
CentOS 5.2
PowerEdge 2970 (2) Quad Core Opteron 2344HE, 4x512K Cache 1.7GHz, 1Ghz
HyperTransport

I have given 5 VCPUs and 4 gigs of RAM to the MSSQL Server, I was planning
on increasing that within the next week.

The storage backend is via iSCSI to an Equallogic PS5000.

Since this is a brand new MSSQL installation (and we have never had one
before), I do not know what the load will look like for us.

I have been noticing some network performance issues in early testing and
will be implementing some changes to the severs this weekend to see if those
are resolved. I will also be looking into the GPLPV which have been
suggested!

Thanks again


Dustin Henning wrote:
> 
> Another important note regarding using Windows on Xen:
> I am pretty sure that using multiple CPUs causes severe performance loss
> with the GPLPV drivers provided by James Harper (to the extent that having
> one VCPU with the drivers may net better performance than having
> multiple).
> However, I believe the GPLPV drivers still provide better performance with
> multiple vcpus than not having the drivers would.  This information could
> be
> slightly dated, but I believe it was recently mentioned by someone else on
> the list.  Regardless, some testing of your scenario options would be duly
> diligent.
>       Dustin
> 

-- 
View this message in context: 
http://www.nabble.com/MSSQL-and-Xen-performance-tp23435695p23448294.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
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