Ujang > i've tried these parameters, but don't help.... > *._gby_hash_aggregation_enabled=FALSE > *._old_connect_by_enabled=TRUE > *._optimizer_cost_based_transformation=OFF > > but when combined those 3 hidden parameter with > *.optimizer_features_enable=8.1.7, it runs so..... fast. I strongly suggest you to understand what your problem is instead of shooting in every direction! E.g. does the execution plan contain a hash group by? If no, why trying the parameter _gby_hash_aggregation_enabled? Same story for _old_connect_by_enabled and _optimizer_cost_based_transformation. If these features are used there are strong evidences in the execution plan. > now, i'm not checking the 10053 trace. by OFE=817, the > v$session_longops give none of information, so i could not > monitor hows the jobs was done.... :( I'm not sure what you are looking in v$session_longops... But since the view does not provide execution plans it is useless for your specific problem. Regards, Chris -- //www.freelists.org/webpage/oracle-l