[TriLUG] Need opinions for specs of a VM box

[David McDowell]

In these tests, were other VMs also running on the same host or you
just virtualized the one DB with no other VMs?  I ask b/c if there
were no other VMs running, what's the point except to see you got 15%
degredation while nothing else was ultilizing that hardware, and then
still proceed with it?

thanks,
David

On 7/27/07, Ron Joffe <rjoffe at yahoo.com> wrote:
>
> On Thursday 26 July 2007 17:49, David McDowell wrote:
> > Your Oracle performs OK in a VM?  what VM host are you using and what
> > size is your DB and how do you measure that performance?  I ask b/c
> > there are a bunch of LUG'ers who are swearing no to DB's in a VM, but
> > with our small MSSQL implementation, I'm not so sure it's gonna be a
> > big deal.
> >
> > David
>
> David,
>
> Our tests have shown the following:
>
> Take a given server (similar spec to what I posted in my previous note, but
> with more drives), Install Oracle on it.
>
> Take an identical server, install vmware, and then install oracle in that vm.
>
> Compare the processing time utilizing an internally developed test (basically
> running a relatively CPU and Disk intensive data conversion job). We have
> seen on the order of 15% performance degradation in the VM versus the real
> host.
>
> A few items to note. Our test only stresses one cpu at a time (i.e. is not
> utilizing oracle parallel operations (threading). And make sure you create
> the same virtual disks to mirror physical disk sets. Base and VM OS is SLES.
>
> In summary, in our environment, we see a 15% loss of performance for moving
> oracle from the base OS to a VM. In our environments, we will typically put
> production Oracle instances on the base OS, and put test and training Oracle
> instances on VM's.
>
> Ron
>
> --
> TriLUG mailing list        : http://www.trilug.org/mailman/listinfo/trilug
> TriLUG Organizational FAQ  : http://trilug.org/faq/
> TriLUG Member Services FAQ : http://members.trilug.org/services_faq/
>