[ovirt-users] oVIRT 3.5 / CPU / Virtual Machines

Yaniv Kaul ykaul at redhat.com
Wed Mar 1 09:52:45 UTC 2017


On Wed, Mar 1, 2017 at 12:38 AM Devin Acosta <devin at pabstatencio.com> wrote:

>
> I am running oVIRT 3.5.6 right now on Dell Blade Chassis, each oVIRT node
> has 2 x 10 core processors, showing 40 VCPU cause of hyper-threading. We
> are trying to figure out for a very log intensive application that requires
> lots of specs what the optimal configuration would be to allow it to use
> utilize all cores on the blade?  Is there advise on this? Like should we
> disable hyperthreading? What's the general rule of thumb?
>

1. Upgrade to the latest oVirt - which provides support for the latest and
greatest hypervisor features.
2. What is your bottleneck - CPU, storage, network, some of each? Many
times, careful NUMA pinning is a good start - after you've ensured the rest
is OK (use of virtio, enough RAM and CPU per VM, storage and network
configurations, etc.). Also - some applications benefit from
hyper-threading, some don't - experiment with both settings.
Y.


> --
>
> Devin Acosta
> Red Hat Certified Architect, LinuxStack
> 602-354-1220 <(602)%20354-1220> || devin at linuxguru.co
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20170301/da3fb127/attachment.html>


More information about the Users mailing list