I have this question running in the MSSQL area also.
I have a running argument with my boss at the moment regarding what RAID set up should be on our new SQL cluster solution.
The current database has two data files on a single filegroup which are heavy in both reads and writes.
The shared storage allows us to allocate 13 disks to this task. I have choosen to create two RAID 10 arrays (4 disks each) for the data files and split them between these two and one RAID 1 array for the transaction logs. The remaining disks are used for the quorum drive and global hot spare.
My boss however is querying my logic and suggests removing the two 4 disk RAID 10 arrays and creating one RAID 10 array with 8 disks and placing all the datafiles on the one virtual disk.
I can see his point, however I would like to get some professional opinions on this. Basically what is better? 2 (4 x disk RAID 10 array) or 1 (8 x disk RAID 10 array). If someone can give me a definate answer I will give them all the points from both questions.