Is this hardware issue?

We are moving our stuff from old hardware to a new hardware. I am observing these results in our new test hardware env: we run a query in the new test env, it runs slow first time it is run. It takes about 6/8 minutes. Once the query is run once, if it is ran again, it comes with results in matter of seconds. This happens a few times and then I flush the buffer cache or bounce the DB and rerun the query, the response time goes back to 7/8 minutes and the subsequent responses are in matter of seconds. I was able to repeat this behavior a few days. I ran the 10046 and checked the plans. The plans in both cases are same. The increased wait seemed to come from waiting longer for the same "db file sequential read" event. Could this point to hardware issues, like I/O or VIO?

No other user is using the database when I do this test, v10.2. OS is AIX, disks on Clarion SAN. Although there are couple of other databases running in this test server, there is no load from them in terms of CPU or memory. I tested in during normal hours, after hours when no one uses any of the DBs in the server. The results are consistent.

--

Thanks,
Ram.

--

http://www.freelists.org/webpage/oracle-l Received on Mon Aug 22 2011 - 14:13:43 CDT