Re: cache buffer chains/where in code
Date: Fri, 20 Nov 2009 14:08:13 +0800
Well that's what i meant by "cpu *starvation*" ;-)
but yeah I get what you mean - it is actually* *possible to get significant queuing for CPU resource without being 100% utilized over multisecond utilization averages. One case I've seen is where a queueing/messaging system invoked bursts of activity by distributing a message to many processes at a time, causing all of them to wake up and do their short work at the same time. Using less connections to the database and bundling more messages into a request solved that issue...
On Fri, Nov 20, 2009 at 1:59 PM, Greg Rahn <greg_at_structureddata.org> wrote:
> To add further color, latch contention can be a symptom of long CPU
> run queues, not just high processor utilization.
> Often times its the waiting (and holding) that is problematic.
> On Thu, Nov 19, 2009 at 8:52 PM, Tanel Poder <tanel_at_poderc.com> wrote:
> > Note that latch contention can also just be a symptom of CPU starvation,
> > if you're running at max physical CPU capacity you should fix that first.
> Greg Rahn
-- Tanel Poder http://blog.tanelpoder.com -- http://www.freelists.org/webpage/oracle-lReceived on Fri Nov 20 2009 - 00:08:13 CST