RE: wierd runstats

From: Cary Millsap <cary.millsap_at_hotsos.com>
Date: Tue, 23 Mar 2004 08:50:22 -0600
Message-ID: <007d01c410e6$2f713f90$1efea8c0@CVMLAP02>

I've seen this pretty frequently. I think if you were to run N concurrent copies of each program (choose N greater than or equal to your system's number of CPUs), you'll probably find that the performance of whatever you did in run2 will degrade much more gracefully. In other words, I would expect the query in run1 to execute much more poorly under load.

Cary Millsap
Hotsos Enterprises, Ltd.
http://www.hotsos.com
* Nullius in verba *

Upcoming events:
- Performance Diagnosis 101: 3/23 Park City, 4/6 Seattle, 5/7 Dallas

SQL Optimization 101: 3/29 Dallas, 4/19 Denver, 5/3 Boston, 5/24 San Diego
Hotsos Symposium 2005: March 6-10 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 ryan.gaffuri_at_cox.net Sent: Tuesday, March 23, 2004 8:41 AM
To: oracle-l_at_freelists.org
Subject: wierd runstats_pkg report

anyone ever see this much of a difference in latch contention and have the query which uses more resources return faster? All I did was flip the join order. I can re-write the query and post it later.

just curious if this has happened to anyone.

SQL> exec runstats_pkg.rs_stop(500);
Run1 ran in 1432 hsecs
Run2 ran in 1740 hsecs
run 1 ran in 82.3% of the time

Name                                Run1      Run2      Diff
STAT...buffer is not pinned co     2,233     1,570      -663
LATCH.library cache pin alloca     1,095       233      -862
LATCH.library cache pin            1,784       347    -1,437
STAT...buffer is pinned count      3,303     1,768    -1,535
STAT...index fetch by key          1,571         3    -1,568
LATCH.shared pool                  2,598       570    -2,028
LATCH.library cache                3,423       835    -2,588
STAT...consistent gets - exami     3,406       144    -3,262
STAT...recursive calls             3,709       406    -3,303
STAT...consistent gets             5,502     1,787    -3,715
STAT...session logical reads       6,049     2,334    -3,715
LATCH.cache buffers chains        10,692     6,384    -4,308
STAT...session uga memory          7,936         0    -7,936

Run1 latches total versus runs -- difference and pct
Run1      Run2      Diff     Pct

23,389 11,135 -12,254 210.05%

PL/SQL procedure successfully completed.

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
-----------------------------------------------------------------

----------------------------------------------------------------
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 Mar 23 2004 - 08:54:57 CST

RE: wierd runstats_pkg report