I'm running a test case for one of our products that's currently in
development and I can't figure out how to account for the total CPU time
for a Java program that executes a DB2 query.
I've got four steps running the same query in different programming
languages. The step statistics are like this
CPU (Total) Elapsed CPU
(TCB) CPU (SRB) Service
Jobname Stepname ProcStep RC I/O hh:mm:ss.th hh:mm:ss.th
hh:mm:ss.th hh:mm:ss.th Units
BENCHSQL C 00 165 00.21 00.38
00.21 00.00 4628
BENCHSQL JAVA SPAWN 00 17049 00.23 02.64
00.23 00.00 34448
BENCHSQL LUA 00 724 00.32 00.88
00.32 00.00 7170
BENCHSQL REXX 00 39 05.08 08.03
05.08 00.00 113K
Java CPU time is very good but the elapsed time is high. I know this is
because of the overhead of the JVM doing a JIT compile but I want to be
able to include that in the report. What SMF
records report Java JVM stats and how do I correlate that to a job step?
I've got a screen shot that shows the numbers
http://imagebin.org/273074. The 1E codes are actually SMF 30s, we
haven't got round to converting them to decimal yet. The 003 codes
are DB2 trace records. The DB2 trace record for the Java program seems
very low. It changes depending on the JDBC driver (2 or 4) but the job
step times remain the same.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
