Hi, sorry for the waiting time, but there has been some mess around. 조동욱 schrieb: > 10046 trace(above level 8) and/or 10053 trace would be a good start point. > If nothing valuable is found by comparison, next step would be searching > metalink. There IS a difference in tracing on events 10046 and 10053 level 8 each between shared and dedicated. So far: Good hint! In dedicated (where the statement(s) are slow) there are lots of wait events "fb contention". What can I do about that? Regards Martin -- Usn's IT Blog for Linux, Oracle, Asterisk www.usn-it.de -- //www.freelists.org/webpage/oracle-l