Resending due to message size. I can forward the plans and/or trace file if anyone is interested. > _____________________________________________ > From: Schultz, Charles > Sent: Tuesday, July 25, 2006 12:39 PM > To: oracle-l@xxxxxxxxxxxxx > Subject: Query tuning exercise: what to look for in a 10053 trace > > I have a query that is going bad when new stats are gathered on the > underlying tables. I checked the stats, and they seem accurate. The > interesting part is that the query plans are quite similar - the > access paths are all exactly the same, but the ordering of the plan > varies a little. So putting on my sql-tuning-expert-wannabe hat, where > is a good place to start? I have not often looked at a 10053 trace to > figure out the order of a plan. From what little I know, I believe it > comes down to join cardinality, but this is an area I have not had > much practice with. I am reading JL's CBO bible ("Cost-Based Oracle > Fundamentals") in an effort to learn a little more. > > Any advice would be appreciated. > Thanks, > > charles schultz > oracle dba > aits - adsd > university of illinois > >