Are you sure that no changes in execution plan occurred during the slowdown? You can easily verify this by querying AWR view :PLAN_HASH_VALUE.dba_hist_sql_plan Of course you have to identify SQL_ID in advance. Michael On Fri, Oct 17, 2008 at 8:03 PM, Dennis Williams < oracledba.williams@xxxxxxxxx> wrote: > Younus > > Does your query use bind variables? I wonder if the first time this SQL is > executed that query is submitted with an unusual value, which causes an > execution plan that uses a FTS. Subsequent executions will continue to use > that execution plan. Your flushing the shared pool causes that execution > plan to be flushed so the next execution gets the execution plan that you > desire. > > Dennis Williams > -- Best Regards Michael Elkin