I have a server with three directly attached disk controllers. Each of these can provide internal redundancy and IO-striping accross disks, but I'm wondering what is the best way to configure the whole system.
Would it be sensible to allocate three datafiles for each tablespace, each datafile being created in a mount provided by a single controller? Oracle seems to distribute data somewhat evenly accross datafiles when doing inserts, but would this be enough striping for queries?
We are mostly concerned about read performance for single aggregate queries (data warehouse) and want to read data in parallel from each controller.
One answer would be to use ASM or LVM to do the striping, Oracle would then see only one mount point where to store datafiles. We do have a configuration with ASM + number of JBOD-disks; it does perform reasonably well, but ASM has it's own issues and we are considering alternatives.