Well.. the 10g docs sort of leave it to the user to figure out what to do with Idle Wait Events.
Here's what the Perf Tuning Guide says :
"The majority of the idle events should be ignored when tuning, because they do not indicate the
nature of the performance bottleneck. Some idle events can be useful in indicating what the
bottleneck is not." citing SQL*Net message from client as an example of the latter.
In 10g there are 58 events in the "Idle" wait class..
I haven't had time to figure out if one can modify this classification to include/exclude certain
events (something I always did with the PERFSTAT.STAS$IDLE_EVENT table ;)
And as for the E2E tracing, one must instrument the appl code to reap benefits of this fantastic
enhancement.
Come to the Hotsos Symposium next month.. there are a couple of presentations discussing 10g OWI
and Tuning stuff...
- Kirti
- Cary Millsap <cary.millsap_at_hotsos.com> wrote:
> > Cary, there are 808 events in 10g.
>
> Thanks.
>
> > There is a huge class of events which oracle calls "idle" and in your
> book you warned against disregarding "idle" events.
>
> I know; so far, I think the whole "wait class" thing is a waste of
> effort--not just the idle stuff, but the whole idea of families of wait
> events. Either a wait event consumes someone's response time, or it
> doesn't. Calling an event a name that makes people feel like it's okay
> to ignore it isn't helping anyone.
>
> Now, there's a difference between partitioning events into families and
> continuing to tell people that it's okay to ignore certain events. I
> haven't checked the 10g doc yet for this; is Oracle Corp. still telling
> people in its documentation to ignore "idle" events?
>
> If they are, then as a technician, I find the idea reprehensible. As an
> owner of a business, I love it. It means that there's a huge class of
> performance problems out there that other tools and methods will still
> fail to find.
>
> > Are there any changes in 10g in that recommendation?
>
> None at all.
>
> > From what I've seen so far, 10g is not so drastically different from
> 9i as 9i was different from 8i. Would you agree with that impression?
>
> I don't think so. The end-to-end tracing model, for example, is a huge
> deal. The difference between 10g instrumentation and 9i instrumentation
> is on the order of the difference between 9i and version 6. The key is
> that "is tracing" is no longer an attribute of an Oracle session; it's
> an attribute of a service, a module, or a [user] action, which is
> exactly where it needs to be! Moving this in the kernel had to be a huge
> headache.
>
>
> Cary Millsap
> Hotsos Enterprises, Ltd.
> http://www.hotsos.com
> * Nullius in verba *
>
> Upcoming events:
> - Performance Diagnosis 101: 2/24 San Diego, 3/23 Park City, 4/6 Seattle
> - Hotsos Symposium 2004: March 7-10 Dallas
> - Visit www.hotsos.com for schedule details...
>
>
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org
> [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Mladen Gogala
> Sent: Tuesday, February 24, 2004 6:03 AM
> To: oracle-l_at_freelists.org
> Subject: Re: 10g query
>
> Cary, there are 808 events in 10g. There is a huge class of events
> which oracle calls "idle" and in your book you warned against
> disregarding
> "idle" events. Are there any changes in 10g in that recommendation?
> >From what I've seen so far, 10g is not so drastically different from
> 9i as 9i was different from 8i. Would you agree with that impression?
>
[snip]
Do you Yahoo!?
Yahoo! Mail SpamGuard - Read only the mail you want.
http://antispam.yahoo.com/tools
Please see the official ORACLE-L FAQ:
http://www.orafaq.com
To unsubscribe send email to: oracle-l-request_at_freelists.org
put 'unsubscribe' in the subject line.
--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Received on Tue Feb 24 2004 - 12:00:11 CST
