When you are working with RAID-0 volumes, consider the following:
Use components that are each on different controllers to increase the number of simultaneous reads and writes that can be performed.
Do not create a stripe from an existing file system or data. Doing so will destroy data. Instead, use a concatenation. (You can create a stripe from existing data, but you must dump and restore the data to the volume.)
Use disk components of the same size for stripes. Striping components of different sizes results in wasted disk space.
Set up a stripe's interlace value to better match the I/O requests made by the system or applications.
Because a stripe or concatenation does not contain replicated data, when such a volume has a component failure, you must replace the component, recreate the stripe or concatenation, and restore data from a backup.
When you recreate a stripe or concatenation, use a replacement component that has at least the same size as the failed component.
Concatenation uses less CPU cycles than striping and performs well for small random I/O and for even I/O distribution.
When possible, distribute the components of a stripe or concatenation across different controllers and busses. Using stripes that are each on different controllers increases the number of simultaneous reads and writes that can be performed.
If a stripe is defined on a failing controller and another controller is available on the system, you can “move” the stripe to the new controller by moving the disks to the controller and redefining the stripe.
Number of stripes: Another way of looking at striping is to first determine the performance requirements. For example, you might need 10.4 Mbytes/sec performance for a selected application, and each disk might deliver approximately 4 Mbyte/sec. Based on this formula, then determine how many disk spindles you need to stripe across:
10.4 Mbyte/sec / 4 Mbyte/sec = 2.6
Therefore, you need three disks capable of performing I/O operations in parallel.