One thing to definetely be aware of is the timed_os_statistics, that you implicity set to 5 per default if setting timed_statistics to ALL. See http://download.oracle.com/docs/cd/B19306_01/server.102/b14237/initparams215.htm#CHDCDEIC Cheers Stefan On Tue, Sep 23, 2008 at 6:16 PM, Hemant K Chitale <hkchital@xxxxxxxxxxxxxx> wrote: > > I have a vendor that insists on running a batch job with > statistics_level='ALL' and event 10046, level 8 ,both set at the session > level in *production* only to diagnose performance issues. Furthermore, it > would seem that the vendor will rely on AWR reports and tkprof's of the > 10046 trace files sent to him post-facto (note : not the trace file, but > the tkprof of it). > > I have run a few tests with my own dummy tables and SQLs to show the impact > of 'ALL' and level 8 to the job execution time. > > I have proposed a Cost/Benefit analysis -- the "Cost" is the increase in job > runtime (for a batch job that can take 8-12 hours normally) and the > "Benefit" being the additional performance diagnostics that is done from the > tkprof's and/or AWRs. > > Do we have a documented / evidenced list of benefits in tkprof when > statistics_level is set to 'ALL' ? > What additional information can I see in an AWR report (generated after the > fact) ? > > > > Hemant K Chitale > > http://hemantoracledba.blogspot.com > > > > -- > //www.freelists.org/webpage/oracle-l > > > -- ========================= Stefan P Knecht Senior Consultant Infrastructure Managed Services Trivadis AG Europa-Strasse 5 CH-8152 Glattbrugg Phone +41-44-808 70 20 Fax +41-808 70 12 Mobile +41-79-571 36 27 stefan.knecht@xxxxxxxxxxxx http://www.trivadis.com OCP 9i/10g SCSA SCNA ========================= -- //www.freelists.org/webpage/oracle-l