Re: High CPU usage

Valentin Minzatu wrote:
> CURSOR_SHARING is a dynamic parameter. If altering it did not provide
> any improvement your problem is not related to the use of literals.
> Post an AWR report in order to receive further help.
>
> BTW flushing the shared pool had adverse effects as it forced *all* the
> queries to be hard re-parsed - not the best idea for an overloaded
> system.

It may or may not be a good idea depending on what _else_ is going on in the shared pool. If there are things that have fragmented or lots of object invalidation has been going on it may help. Since the OP said he waiting 5 hours for the system to normalize, I think he understands that. Flushing it as a matter of course would be another matter.

What isn't so clear is whether high CPU usage is a problem, and that isn't going to be clear, since the CPU's aren't spinning at 100%. From the thread so far, it doesn't even appear that there is a problem, and several potential problems have been ruled out. Perhaps the OP can restate what he thinks the problem is? Simply saying the cpu usage used to be 20% and now is 70% may simply mean the old system was more i/o bound or otherwise _less_ efficient.

I can speculate the newer configuration is spending more time thrashing the SGA, I'd say perhaps a look at V$BH (see queries in the performance tuning guide, like
http://download-west.oracle.com/docs/cd/B19306_01/server.102/b14211/memory.htm ) might be informative.

jg

--
@home.com is bogus.
http://www.signonsandiego.com/teaserimages/061013airshow_b.jpg