Identifying segment experiencing specific wait event

buffer busy waits event occurred 3400 times.In order to tune the buffer busy waits, we could do a lot of

things — we could increase the freelist groups and freelists of the segment, or we could rearrange the distribution of the rows in the table in such a way that the blocks are not repeatedly

picked up at the same time from two different session. However, to do either of these, it's necessary to know the exact segment that to be tuned. The STATSPACK report does not tell us which objects contributed to the buffer busy waits event, and without the segment level information, the tuning cannot

possibly continue. Traditionally, we would place event 10046 for each of the sessions and see all the wait events in the generated trace files, which tend to be extremely large. In a typical system, which may contain several hundred applications, this approach may not be feasible. Additionally if the applications connect through Multi Threaded Server, it becomes difficult to isolate single segment level problems even

if trace analysis is possible. This information is now obtained from the new performance

view v$segstat and v$segment_statistics.