Vivek
Your statspack report detail says that the bulk of the (not much) work carried out in the period was contributed by statspack itself.
7 wait seconds (across 25 application users) out of 7 minutes, and 12 cpu seconds used out of a possible 7 * 60 * 8 = 3360 (the statspack says you have 8 cpus) is not worth wasting your time over. Just make sure that your work is well balanced across the RAC instances (not necessarily symmetrically). This load looks much lighter than the other example you have circulated (with gc buffer busy waits). And check statspack isn't being misled - eg if some statements were incomplete at the end of the period, their costs wouldn't be accumulated into the statspack views. Look at a report from snap 27 to 29, if you have one. And don't make the statspack periods too short in future ...
Regards Nigel
--
http://www.freelists.org/webpage/oracle-l
Received on Tue Feb 27 2007 - 11:48:22 CST
