Comments on: 10046 Extended SQL Trace Interpretation 3 http://hoopercharles.wordpress.com/2010/09/06/10046-extended-sql-trace-interpretation-3/ Miscellaneous Random Oracle Topics: Stop, Think, ... Understand Mon, 13 May 2013 14:10:06 +0000 hourly 1 http://wordpress.com/ By: Pavel Ermakov http://hoopercharles.wordpress.com/2010/09/06/10046-extended-sql-trace-interpretation-3/#comment-1828 Mon, 13 Sep 2010 09:07:44 +0000 http://hoopercharles.wordpress.com/?p=3201#comment-1828 Hi Charles!

Thanks for the response.

Best regards, Pavel.

]]> By: Charles Hooper http://hoopercharles.wordpress.com/2010/09/06/10046-extended-sql-trace-interpretation-3/#comment-1821 Fri, 10 Sep 2010 13:00:48 +0000 http://hoopercharles.wordpress.com/?p=3201#comment-1821 Pavel,

Good question regarding the two executions in the script of:

EXEC DBMS_STATS.GATHER_TABLE_STATS(OWNNAME=>USER,TABNAME=>'T3_1',CASCADE=>TRUE)
EXEC DBMS_STATS.GATHER_TABLE_STATS(OWNNAME=>USER,TABNAME=>'T4_1',CASCADE=>TRUE)

The second execution of the statistics gathering was NOT performed to change the execution plan. The NO_INVALIDATE parameter (of DBMS_STATS.GATHER_TABLE_STATS) was not set to FALSE, so on Oracle Database 10.1 and above those statistics would not have immediately affected previously generated execution plans.

I selected the statistics gathering lines as the starting point where the script repeated, so that the timings in both 10046 trace files would be affected roughly the same by any data that might remain in the hard drive caches when Oracle’s buffer cache was flushed. It might be interesting to see whether or not removing those lines will affect the outcome of the test results.

]]> By: Pavel Ermakov http://hoopercharles.wordpress.com/2010/09/06/10046-extended-sql-trace-interpretation-3/#comment-1819 Fri, 10 Sep 2010 11:34:25 +0000 http://hoopercharles.wordpress.com/?p=3201#comment-1819 Hi Charles!

I have a question: why did you gather statistics for the tables twice?

]]>