HP Surestore Disk Array 12h - And FC60, Advanced User'S Manual, Page: 62 / 466

62
Disk Array High Availability Features
Settings that give a higher priority to the rebuild process will cause the rebuild to complete
sooner, but at the expense of I/O performance. Lower rebuild priority settings favors host
I/Os, which will maintain I/O performance but delay the completion of the rebuild.
The rebuild priority settings selected reflect the importance of performance versus data
availability. The LUN being rebuilt is vulnerable to another disk failure while the rebuild is
in progress. The longer the rebuild takes, the greater the chance of another disk failure.
The following sequence occurs following a disk failure and replacement. Figure 23
illustrates the process. A 5-disk RAID 5 LUN is used for this example.
1. Disk 3 in the RAID 5 LUN fails.
2. The disk array locates an available global hot spare and begins recreating on it the
information that was on the failed disk. The data and parity on the remaining four disks
in the LUN are used to recreate the information.
3. When the rebuild finishes, the global hot spare is part of the LUN, which fulfills the roll
of disk 3.
4. When disk 3 is replaced, the disk array begins copying all the information from the
former global hot spare to the replacement disk.
5. When copying completes, the LUN is restored to its original configuration. The former
global hot spare is returned to the global hot spare disk group and is available to protect
against another data disk failure.
Note
Can a lower capacity disk serve as a hot spare for a larger disk?
It is possible for a lower capacity disk to be used as a global hot spare when a
larger disk fails. When a disk failure occurs, the disk array controller looks for
a global hot spare that is large enough to store the data on the failed disk, not
for a disk that matches the capacity of the failed disk. For example, if an 18
Gbyte disk fails but there is only 6 Gbytes of data stored on the disk, a 9 Gbyte
global hot spare could be used.
Although this feature is available, it is recommended that you always select the
largest disks in the array to serve as global hot spares. This will ensure that any
disk in the array is protected, regardless of how much data is stored on it.