What do you make of this?

From: Paul Baumgartel <paul.baumgartel_at_gmail.com>
Date: Fri, 9 Sep 2005 12:18:28 -0400
Message-ID: <f8c477105090909186deb62bf@mail.gmail.com>

Hi, all. Oracle10g (10.1.0.4.0) on Windows 2000.

I have a set of traces that are puzzling. I had the traces done in controlled conditions, with a single user on the database; CPU usage on the DB server was less than 25%. This statement is a good example of what I'm trying to figure out. There's a high ratio of elapsed to CPU on the parse, but not much in the way of wait information--except the SQL*Net message from client. The client is a COM+ application server. No recent changes that I know of on that machine, nor on the DB server itself. I have inspected the raw trace file as well and don't see anything to suggest that the tkprof timing summaries are incorrect.

II'm looking for ideas on how to go about diagnosing the SQL*Net waits (assuming they're accounting for the different between CPU and elapsed) or finding where the rest of the elapsed time comes from. Thanks in advance for any suggestions!

UPDATE VALIDKEYS SET LASTACCESS = :V00001 WHERE
VALIDKEYS_PK = :V00002 call count cpu elapsed disk query current rows
------- ------ -------- ---------- ---------- ---------- ----------

Parse 198 0.46 2.72 0 0 0 0
Execute 99 0.12 0.49 0 297 396 99
Fetch 0 0.00 0.00 0 0 0 0
------- ------ -------- ---------- ---------- ---------- ----------

total 297 0.59 3.22 0 297 396 99

Misses in library cache during parse: 0
Optimizer mode: ALL_ROWS
Parsing user id: 12492

Elapsed times include waiting on following events: Event waited on Times Max. Wait Total Waited

Waited ---------- ------------ SQL*Net message to client 381 0.00 0.00 SQL*Net message from client 381 0.01 0.51 log file sync 1 0.00 0.00

--

Paul Baumgartel
paul.baumgartel_at_aya.yale.edu

--

http://www.freelists.org/webpage/oracle-l Received on Fri Sep 09 2005 - 11:20:40 CDT