Re: Was simple performance stats between Xen 2.0 and 3.0 (was Re: [Xen-users] Xen, NFS performance, rsize, wsize and MTU)

[Ernst Bachmann <e.bachmann@xxxxxxxx>]

On Wednesday 22 February 2006 20:09, Nicholas Lee wrote:
> On 23/02/06, Chris Fanning <christopher.fanning@xxxxxxxxx> wrote:
> > I was trying to set up home accounts on a domU.
> > I've tested what you said about homes on domU and guess what, you're
> > right
> >
> > :)
> >
> > I'm going to set them up on dom0 (or somewhere else).
>
> I've got two machines in a colo now with Xen. Now I've sort out my bridging
> problem I can do a bit of simple testing. Its a reasonable difference.
>
> >From x335 domU (running 2.0) to v20z (running 3.0):
[...]
>
> Still there is a large difference between dom0 to domU and domU to domU and
> obviously a domU in Xen 2.0 is not able to handle supplying other domUs on
> the same host0 effectively.

How were the VCPUs distributed among the domains? all domains running on 
single-CPU or hyperthreading or SMP?

If you have SMP/HT available, maybe try to dedicate one thread/core to dom0 
exclusively, otherwise for each IO operation xen has to switch context 
between them.

/Ernst

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