Hi Sinh, Depending on flavor size and benchmark, we were able to reduce the loss compared to bare metal from 8-20% to 3-5%. Details can be found on our blog:
http://openstack-in-production.blogspot.ch/2015/08/numa-and-cpu-pinning-in-high-throughput.html and in more depth in this presentation: https://indico.cern.ch/event/384358/contributions/909247/ Cheers, Arne > On 07 Feb 2017, at 10:06, Sinh Lam <sinh....@verizondigitalmedia.com> wrote: > > hi arne, > > do you happen to have performance numbers before and after creating guests > with and without the cpu pinning options? > > thanks. > > Sinh > > > On Mon, Feb 6, 2017 at 11:46 PM, Arne Wiebalck <arne.wieba...@cern.ch> wrote: > Manuel, > > Rather than with aggregate metadata we assign instances to NUMA nodes via > flavor extra_specs, > i.e. nova flavor-show reports something like > > —> > | extra_specs | {"hw:numa_nodes": "1"} | > <— > > for our NUMA-aware flavors. > > This seems to work pretty well and gives the desired performance improvement. > > Cheers, > Arne > > > > > On 07 Feb 2017, at 01:19, Manuel Sopena Ballesteros > > <manuel...@garvan.org.au> wrote: > > > > Hi, > > > > I am trying to isolate my instances by cpu socket in order to improve my > > NUMA hardware performance. > > > > [root@openstack-dev ~(keystone_admin)]# nova aggregate-set-metadata numa > > pinned=true > > Metadata has been successfully updated for aggregate 1. > > +----+--------+--------------------+--------+--------------------+ > > | Id | Name | Availability Zone | Hosts | Metadata | > > +----+--------+--------------------+--------+--------------------+ > > | 1 | numa | - | | > > 'pinned=true' | > > +----+--------+--------------------+--------+--------------------+ > > > > I have done the changes on the nova metadata but my admin can’t see the > > instances > > > > [root@openstack-dev ~(keystone_admin)]# nova aggregate-add-host numa > > 4d4f3c3f-2894-4244-b74c-2c479e296ff8 > > ERROR (NotFound): Compute host 4d4f3c3f-2894-4244-b74c-2c479e296ff8 could > > not be found. (HTTP 404) (Request-ID: > > req-286985d8-d6ce-429e-b234-dd5eac5ad62e) > > > > And the user who has access to those instances does not have privileges to > > add the hosts > > > > [root@openstack-dev ~(keystone_myuser)]# nova aggregate-add-host numa > > 4d4f3c3f-2894-4244-b74c-2c479e296ff8 > > ERROR (Forbidden): Policy doesn't allow os_compute_api:os-aggregates:index > > to be performed. (HTTP 403) (Request-ID: > > req-a5687fd4-c00d-4b64-af9e-bd5a82eb99c1) > > > > What would be the recommended way to do this? > > > > Thank you very much > > > > Manuel Sopena Ballesteros | Big data Engineer > > Garvan Institute of Medical Research > > The Kinghorn Cancer Centre, 370 Victoria Street, Darlinghurst, NSW 2010 > > T: + 61 (0)2 9355 5760 | F: +61 (0)2 9295 8507 | E: manuel...@garvan.org.au > > > > NOTICE > > Please consider the environment before printing this email. This message > > and any attachments are intended for the addressee named and may contain > > legally privileged/confidential/copyright information. If you are not the > > intended recipient, you should not read, use, disclose, copy or distribute > > this communication. If you have received this message in error please > > notify us at once by return email and then delete both messages. We accept > > no liability for the distribution of viruses or similar in electronic > > communications. This notice should not be removed. > > _______________________________________________ > > Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > > Post to : openstack@lists.openstack.org > > Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > > -- > Arne Wiebalck > CERN IT > > _______________________________________________ > Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > Post to : openstack@lists.openstack.org > Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > -- Arne Wiebalck CERN IT _______________________________________________ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
