The level of mirrored protection determines whether the system keeps running when different levels of hardware fail. Mirrored protection always provides disk-unit level protection, which keeps the system available for a single disk-unit failure. To keep the system available for failures of other disk-related hardware requires higher levels of protection. For example, to keep the system available when an I/O processor (IOP) fails, all of the disk units attached to the failing IOP must have mirrored units attached to different IOPs.
The level of mirrored protection also determines whether concurrent maintenance can be done for different types of failures. Certain types of failures require concurrent maintenance to diagnose hardware levels above the failing hardware component. For example, to diagnose a power failure in a disk unit, you need to reset the I/O processor to which the failed disk unit is attached. Therefore, IOP-level protection is required. The higher the level of mirrored protection, the more often concurrent maintenance is possible.
The level of protection you get depends on the hardware you duplicate. If you duplicate disk units, you have disk-unit level protection. If you duplicate IOAs as well, you have IOA-level protection. If you duplicate input/output processors, you have IOP-level protection. If you duplicate buses, you have bus-level protection. Mirrored units always have at least disk-unit level protection.
During the start mirrored protection operation, the system pairs the disk units to provide the maximum level of protection for the system. When disk units are added to a mirrored disk pool, the system pairs only those disk units that are added without rearranging the existing pairs. The hardware configuration includes both the hardware and how the hardware is connected.