Re: Interpreting statspack report

From: Sybrand Bakker <postbus_at_sybrandb.demon.nl>
Date: Wed, 23 Nov 2005 21:44:02 +0100
Message-ID: <uvk9o1t6lvs10p5ndamg0a82vcs87g7f30@4ax.com>

On 23 Nov 2005 12:15:14 -0800, "EdStevens" <quetico_man_at_yahoo.com> wrote:

>Do we have a problem or are we tuning for the sake of tuning?
>
>Maybe I'm missing something, but I'd think most people would be very
>happy with 85% of their time in the CPU. Most people are trying to
>wring out their wait events.

Sorry to disagree. I'm currently addressing a situation when then end-user spends 95 percent of their time in the CPU, and is complaining about performance. And the end-user is right, performance is abysmal. Run the affected query (with 4 left outer joins) in ALL_ROWS and you abort it after 46 minutes, run it in first_rows, and you have the complete 20000 records in 19secs.

--
Sybrand Bakker, Senior Oracle DBA

Received on Wed Nov 23 2005 - 14:44:02 CST