Re: Negative ramifications of setting CPU_COUNT lower?
Date: Fri, 6 Nov 2009 08:49:01 -0800
Message-ID: <a9c093440911060849q4d8348f8s4533d38a1eaef5f8_at_mail.gmail.com>
I know of several Sun/Oracle customers who have successfully implemented the CMT stuff and have replace some fairly large systems (15K, 6800/6900, etc) with the new T5540 or similiar.
Part of the problem as I see it is that many people don't seem to understand the trade offs for the CMT processor line. I usually tell people that if they have a throughput based workload (lots of sessions doing lots of short transactions) then it works well. Its akin to having a 4 lane freeway at 80 MPH compared to say a 12 lane freeway at 40. Depending on how many lanes you can keep "busy" with cars, the wider, slower road may move more cars in a fixed time period. It should be fairly obvious then if there is only a few number of cars, then the faster/fewer option is better.
So instead of being scared of something, I think its much better to explain and understand the trade offs and why it is so.
On Fri, Nov 6, 2009 at 8:31 AM, Allen, Brandon
<Brandon.Allen_at_oneneck.com> wrote:
> I have no experience with changing cpu_count, but I do have a little
> experience with running the Oracle database on T-series and I can sum it up
> in two words: “Run away!”
-- Regards, Greg Rahn http://structureddata.org -- http://www.freelists.org/webpage/oracle-lReceived on Fri Nov 06 2009 - 10:49:01 CST