[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] xen: Implement domain runstates


  • To: Bruce Edge <bruce.edge@xxxxxxxxx>, George Dunlap <George.Dunlap@xxxxxxxxxxxxx>
  • From: Keir Fraser <keir@xxxxxxx>
  • Date: Thu, 25 Nov 2010 22:24:54 +0000
  • Cc: "xen-devel@xxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Thu, 25 Nov 2010 14:27:38 -0800
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=sender:user-agent:date:subject:from:to:cc:message-id:thread-topic :thread-index:in-reply-to:mime-version:content-type :content-transfer-encoding; b=tU1CT5y7iLm3uRjkhbgcGzHQ9KkTHFGRlsB9q3Nkq0lPS9je5eYM7CCCoMUEnz3PaU YNiLB1yTqjljFI/JO7SyVCp4IVidk8SniSTfgpavb3kttjL7LD0U1EEHGVN+Gy9gmmn0 ri44pdvvz6seg9NWIosNiU0SvLq7x9wdPpeco=
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>
  • Thread-index: AcuM75VLlnrQxIYB/EqbfjbjdPkhVw==
  • Thread-topic: [Xen-devel] [PATCH] xen: Implement domain runstates

On 25/11/2010 20:02, "Bruce Edge" <bruce.edge@xxxxxxxxx> wrote:

> Irrespective of the nature of the suggested functionality, I do agree that xen
> users will be more likely to try something if it's part of the default build,
> especially if the benefit is not immediately obvious. With all the complexity
> associated with any virtualization project, I think that many integrators
> don't have the time to experiment with every possible option included in the
> "off the shelf" Xen, not to mention those that require explicit patching and
> manual inclusion.
> That said I would try this as a "VCPU allocation suitability" metric for any
> xen system. The ability to see this data 'at a glance' would be beneficial to
> admins and system architects.
> A lot of data is available if you know how to get it. That is very different
> from presenting that data in an easily digestible format requiring no work to
> use.

You may be under the impression the feature is free. It isn't -- it adds a
lock and loop over all vcpus on every vcpu scheduling/descheduling decision.
That could be frequent for VCPUs doing a lot of short blocking, for example.
And the system-wide work done to support this feature will scale
quadratically on average with vcpus-per-domain. My gut feeling is this isn't
a good thing to have on your context switch path.

 -- Keir



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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.