You really can't compare tkprof to statspack in this case. In the sql trace file, you have approximately 33k executions of the statement, in statspack you have 960k (almost 30 times as many). You could have one or more executions in your sql trace that were so far in excess of 2ms that it totally skews the average.
Regards,
Daniel Fink
hi, all
While checking some performance issue on a database
(9205/solaris8), i noticed the SQL performance from
statspack/sql_trace_tkprof_file are different; sql_trace file shows
each execution is about 5.3 buffer_gets and 2ms. but statspack shows
each execution elapsed time is about 0.5ms (too fast, ah)...
call count cpu elapsed disk query current rows
------- ------ -------- ---------- ---------- ---------- ---------- ----------
Parse 0 0.00 0.00 0 0
0 0
Execute 32945 55.52 60.49 671 134208 41604 32945
Fetch 0 0.00 0.00 0 0 0 0
------- ------ -------- ---------- ---------- ---------- ---------- ----------
total 32945 55.52 60.49 671 134208 41604 32945
But statspack shows each execution is about 0.48ms(rounded to 0.5);
CPU Elapsd
Buffer Gets Executions Gets per Exec %Total Time (s) Time (s) Hash Value
--------------- ------------ -------------- ------ -------- --------- ----------
5,008,886 960,368 5.2 30.9 349.27 461.47 1975398672
--
Regards
--
http://www.freelists.org/webpage/oracle-l
Received on Tue Aug 21 2007 - 11:12:26 CDT
