We are using EBS R12.2.9 with Database 19c (19.7). Both Application tier and Database tier(Non-Rac, Non-ASM) running on separate servers. Database size 3.7 TB.
In Database server, We have three mountpoints containing datafiles, one mountpoint for Oracle Binaries, CDB datafiles and redo logs, and one more point to archive logs. All these mountpoints are created using single lun to storage. Now storage team wants to create multiple luns and mount those as 5 mountpoints for datafiles(1 Lun for 1 mountpoint). So we have to redistribute datafiles across 5 mountpoints. we are planningto redistribute all datafiles by using data file_id as below. Kindly provide your inputs on this approach. If you have any suggestions, please let me know.
FYI, We have 160 datafiles for PDB. We had ran ASH report also to identify around 15 hot datafiles and not to keep all those datafiles in one mountpoint. We are shuffling those datafiles across mountpoints.
Then, organize your mount points/disks/raid groups/adapters according to the SAN/Storage best practices.
Consider aligning your OS level disk configuration to the SAN level disk configuration. (we don't want to do 1 IO in 2 IOS. :)
For Redo, use disks optimized for writes.
For DB files, use disks mostly optimized for DB I/O.
As for disk count and mount point count, we don't have a general suggestion.
The important thing is not to be blocked anywhere in the OS I/O subsystems while doing heavy and concurrent DB I/O.. Having a single big physical disk may serialize some I/O request in the OS level.. (in some queues) and we generally don't want that.
Just take your db size as a reference, and try to be logical, while sizing the underlying disks.
For instance, if the db is 2TB, 4 different disk source (each 500 GB) may be a good start. I mean 500 GB usable space for each raid group maybe..
Aha.. Also, configure RAID. We use raid mirror for write intensive disks, we use both fault tolerant and high performance raid configurations for the raid groups where the db files reside.. Acutally, all depend on the storage.. The question is very generic, but I did try my best to answer in this very short time.