cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From mphilli7823 <mphilli7...@hotmail.com>
Subject RE: CS cores vs sockets
Date Thu, 17 Apr 2014 18:36:59 GMT
Would seem to be an easy to do...


Sent via the Samsung GALAXY S®4, an AT&T 4G LTE smartphone

<div>-------- Original message --------</div><div>From: Marcus <shadowsor@gmail.com>
</div><div>Date:04/17/2014  1:09 PM  (GMT-06:00) </div><div>To: dev@cloudstack.apache.org
</div><div>Subject: Re: CS cores vs sockets </div><div>
</div>
Or like you mention, service offerings could have a 'cores per socket' setting.

On Thu, Apr 17, 2014 at 12:06 PM, Marcus <shadowsor@gmail.com> wrote:
> With KVM this has been mitigated somewhat in newer code (4.3 and up).
> It's a bit rudimentary, but we could expand it to work via config
> options. Right now, if the number of cores is divisible by 4, it
> creates quad core sockets. If the number is divisible by 6, it creates
> hexacore sockets, with the hexacore taking precedence for something
> like 12.
>
> On Thu, Apr 17, 2014 at 11:46 AM, Michael Phillips
> <mphilli7823@hotmail.com> wrote:
>> I think this issue may have been raised in the past, but was not addressed..
>> When creating a service offering in CS using multiple "cores" CS, actually creates
the VM in the background with multiple sockets. Example a 6 "core" offering actually translates
into a 6 socket offering. This is a problem on certain OS's and applications like SQL 2012.
SQL 2012 has a 4 socket maximum. The easiest fix, might be just to allow the admin to specify
how to arrive at the core count when creating the service offering.
>> http://msdn.microsoft.com/en-us/library/ms143760.aspx
>> I can't speak for other hypervisors, but this definitely effects vmware. The only
workaround without a fix is to change the vm via virtual center...
>> Thoughts?

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message