From mboxrd@z Thu Jan 1 00:00:00 1970 From: Matt Wilson Subject: Re: [PATCH 0 of 8] NUMA Awareness for the Credit Scheduler Date: Tue, 9 Oct 2012 13:20:19 -0700 Message-ID: <20121009202001.GA24955@u002268147cd4502c336d.ant.amazon.com> References: <1349779549.3610.79.camel@Abyss> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <1349779549.3610.79.camel@Abyss> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Dario Faggioli Cc: Marcus Granado , Dan Magenheimer , Andre Przywara , Ian Campbell , Anil Madhavapeddy , George Dunlap , Andrew Cooper , Juergen Gross , Ian Jackson , xen-devel@lists.xen.org, Jan Beulich , Daniel De Graaf List-Id: xen-devel@lists.xenproject.org On Tue, Oct 09, 2012 at 11:45:49AM +0100, Dario Faggioli wrote: > > Whether, that is acceptable or not, is of course debatable, and we had a > bit of this discussion already (although no real conclusion has been > reached yet). > My take is that, right now, since we do not yet expose any virtual NUMA > topology to the VM itself, the behaviour described above is fine. As > soon as we'll have some guest NUMA awareness, than it might be > worthwhile to try to preserve it, at least to some extent. For what it's worth, under VMware all bets are off if a vNUMA enabled guest is migrated via vMotion. See "Performance Best Practices for VMware vSphere 5.0" [1] page 40. There is also a good deal of information in a paper published by VMware labs on HPC workloads [2] and a blog post on NUMA load balancing [3]. Matt [1] http://www.vmware.com/pdf/Perf_Best_Practices_vSphere5.0.pdf [2] http://labs.vmware.com/publications/performance-evaluation-of-hpc-benchmarks-on-vmwares-esxi-server [3] http://blogs.vmware.com/vsphere/2012/02/vspherenuma-loadbalancing.htmlvnu