This topic provides guidelines about your storage subsystems logical disks.
Most storage subsystems provide some mechanism to create multiple logical disks from a single array. This is useful when the storage subsystem is directly presenting storage to the hosts. In a virtualized SAN, however, where possible, there should be a one-to-one mapping between arrays and logical disks. Ensuring that the arrays are configured in this way will make the subsequent load calculations and the managed disk and managed disk group configuration tasks a lot easier.
For example, you have two RAID-5 arrays and both contain 5 + P components. Array A has a single logical disk that is being presented to the SAN Volume Controller cluster. This is seen by the cluster as mdisk0. Array B has three logical disks that are being presented to the SAN Volume Controller cluster. These are seen by the cluster as managed disks 1 through 3. All four managed disks are assigned to the same managed disk group, mdisk_grp0. When a virtual disk is created by striping across this group, what actually takes place is that array A presents the first extent and array B presents the next 3 extents. Therefore, when reading and writing to the virtual disk, the loading is split 25% on the disks in array A and 75% on the disks in array B. The performance of the virtual disk will in general be one third of what array B can sustain.
This example describes the performance degradation and complexity that is introduced by having uneven logical disks in a simple configuration. As stated in the guideline summary, you should aim to create a single logical disk from each array.