I have an instance that was built for Cognos reports. There are several big tables (1~3G), unfortunately, located on one disk, making the disk hot. My goal is to break these tables up such that they spread across three disks.
That would divide up the file I/O based upon which table lies in which dbf, but the I/O of all the dbf contributes to a high ts I/O in the case that multiple transactions are being carried out on the ts.
On the other hand you can find out the type of activity on these tables like inserts, deletes, read or write intensive, etc., and create 3 ts with preferably one dbf per ts (one ts per disk), with each ts containing the tables which undergo a specific type of activity. I think this would be a better approach. Does anyone agree?