Re: cpu_count per instance in a hard partition

AFAIK, Oracle instances are not aware of other instances running in the same host and, even though, why would it matter to them? They see the CPUs as any other software running on the same server, be it Oracle DBA, WebLogic or the OE itself, they are competing to get the resources, not only CPU but also memory and IO. So, you have five instances running and it is up to you to figure out if they will be struggling to get CPU cycles or they will just come along fine, each instance will assume that it has as many CPUs available as it seems and act accordingly.
Bear in mind that Oracle does not charge you for the number of instances running or the number of different versions, just the cores the server have so, if the instances won't fight too much for CPU, then you are actually saving money. On the other side, if they start struggling you may have to consider adding some additional CPU and licenses.

I hope this helps.

On Fri, 23 Nov 2018, 14:55 Patrice sur GMail <patrice.boivin_at_gmail.com wrote:

> Hi,
>
> a possibly dumb question:
>
> If you have a hard partition with 2 cores but 8 threads total but are
> running five Oracle instances within that hard partition, does the Oracle
> software know that five instances are running from the same home? Or is
> each instance oblivious to the fact that it is sharing resources with other
> instances.
>
> I ask because if we set cpu_count=8 in each of five instances, doesn't
> that mean the instances expect the hard partition to provide a total of 40
> threads?
>
> If they are aware of their environment then each instance could say "well
> obviously they are lying to me, I only really have 8/5 threads available to
> me, though we are sharing resources so sometimes I will have more time
> slices, other times, not as many."
>
> Overloading may not cause major problems other than each instance not
> getting as many time slices as it expects, i.e. each one would end up
> running slower than optimal. (reminds me of when I was a student working
> on an overloaded mainframe which had more dumb terminals than recommended,
> each dumb terminal had to wait for time slices) Meanwhile there would be
> nearly no wasted time slices because the CPUs are very busy.
>
> -- Patrice
> My profiles: [image: Facebook]
> <http://www.facebook.com/home.php?#!/profile.php?id=100000206805521>[image:
> LinkedIn] <http://ca.linkedin.com/pub/patrice-boivin/a/933/5a9>[image:
> Twitter] <http://www.twitter.com/PatriceBoivin>
>

-- 


--






--
http://www.freelists.org/webpage/oracle-l