Re: Oracle performance and swapping on Tru 64

Rob I don't quite understand why you posted the vmstat and info from "during the day" when you say the problem is occurring during the night batch run. So please post the information you have from the night run.

What the unix system admins say does make sense and does not indicate a memory problem.

My initial recommendation here would be to follow a tuning methodology such as documented by Cary Millsap in "Optimizing Oracle Performance".

Select an appropriate time during the batch run to collect properly scoped diagnostic data and slap a 10046 trace on the batch process.

What exactly is slowing down the batch processing?

That is the type of approach that is likely to yield improvements. If your results from diagnostic data analysis do point to memory related problems, fine, then you have evidence that will be useful.

If it points in some other direction then follow where the result of the analysis yields.

Egor S has an excellent free tool "ORASRP" that can be used to look at the 10046 trace data ( a resource profilerI. Or you can use tkprof. Received on Mon Mar 20 2006 - 07:21:28 CST