Understanding Hot Spares Back to Storage Management Contents Page
Dell OpenManage™ Server Administrator Storage Management User's Guide
Considerations for Hot Spares on PERC 3/Si, 3/Di, and CERC SATA1.5/6ch Controllers
When you create a redundant virtual disk using a RAID controller, you have the opportunity to maintain system operations even when a disk fails. To do so, you would assign a hot spare to the virtual disk. When a disk fails, the redundant data is rebuilt onto the hot spare without interrupting system operations.
A hot spare is an unused backup array disk that can be used to rebuild data from a redundant virtual disk. Hot spares remain in standby mode. When an array disk that is used in a redundant virtual disk fails, the assigned hot spare is activated to replace the failed array disk without interrupting the system or requiring your intervention. If a virtual disk using the failed array disk is not redundant, then the data is permanently lost without any method (unless you have a backup) to restore the data.
Hot spare implementation is different for different controllers. See the following sections for more information.
The following sections describe procedures for assigning a hot spare:
On the PERC 3/SC, 3/DCL, 3/DC, 3/QC, 4/SC, 4/DC, 4e/DC, 4/Di, 4e/Si, 4e/Di, CERC ATA100/4ch, PERC 5/E controllers, assigning a hot spare is equivalent to assigning an array disk to replace another array disk if it fails. If more than one redundant virtual disk resides on the array disk, then all redundant portions of the array disk are rebuilt.
 |
Note: When rebuilding an array disk, you need to delete any nonredundant virtual disks (such as RAID 0) that reside on the array disk before rebuilding the array disk. |
When creating a virtual disk, the array disks included in the virtual disk can be different sizes. When assigning a hot spare to a RAID 1 or 5 virtual disk, the hot spare only needs to be the same size (or larger) as the smallest array disk included in the virtual disk.
This is because when using a PERC 3/SC, 3/DCL, 3/DC, 3/QC, 4/SC, 4/DC, 4e/DC, 4/Di, 4e/Si, 4e/Di, CERC ATA100/4ch, PERC 5/E controller, you can assign array disks of different sizes to a virtual disk. When you have fully consumed a smaller array disk with a virtual disk, however, any portion of larger array disks that are not consumed by the virtual disk become unusable. Therefore, there is no data on the unused portion of a larger disk that needs to be rebuilt. A redundant virtual disk will also be either striped or mirrored in equal portions across its member array disks. The amount of data requiring a rebuild will therefore not be larger than the smallest array disk.
A RAID 10 or 50 virtual disk may include spans that have array disks of different sizes. In this case, you should identify the span that has the largest "small" array disk. The hot spare should be large enough to rebuild this array disk. For example, if one span has three array disks that are 60 MB, 60 MB and 40 MB and another span has array disks that are 60 MB, 60 MB, and 50 MB, then the hot spare must be 50 MB or larger.
A dedicated hot spare can only be assigned to the set of virtual disks that share the same array disks. A global hot spare is assigned to all redundant virtual disks on the controller. A global hot spare must be the same size (or larger) as the smallest array disk included in any virtual disk on the controller.
After you have assigned a global hot spare, any new virtual disks created on the controller will not be protected by the hot spare in either of the following circumstances:
In this case, you can unassign the global hot spare after creating a new virtual disk and then assign a new and larger hot spare to cover all redundant virtual disks on the controller. See "RAID Controller Technology: SCSI, SATA, ATA, and SAS" to determine whether the controller is using SCSI or SAS technology.
On the PERC 3/SC, 3/DCL, 3/DC, 3/QC, 4/SC, 4/DC, 4e/DC, 4/Di, 4e/Si, 4e/Di, and CERC ATA100/4ch controllers, the virtual disk state is not updated until the controller performs an I/O operation. This means that when a redundant virtual disk is degraded on one of these controllers, the hot spare will not be activated until the controller performs an I/O operation. See "I/O Requirements for Detecting Drive Status Changes" for more information.
The following considerations apply to dedicated hot spares:
If you have a hot spare assigned to a virtual disk and an array disk in the virtual disk fails, the failed array disk may change from Online state to Ready state without displaying a Failed state. This occurs when the hot spare is activated before the array disk is able to report the Failed state. Because the Failed state is not reported, the "Device failed: Array disk" event "2048" is not generated.
When the hot spare is activated it changes to the Rebuilding state. If you review the event log and identify a "rebuilding" event such as "2064" or "2065", you can assume that an array disk has failed.
For the PERC 3/Si, 3/Di, and CERC SATA1.5/6ch controllers, a hot spare is assigned to a virtual disk. When an array disk fails, only the portion of the array disk containing the virtual disk is rebuilt onto the hot spare. Data or space on the array disk not included in the virtual disk will not be rebuilt.
On the PERC 3/Si, 3/Di, and CERC SATA1.5/6ch controllers, individual array disks may be included in more than one virtual disk. (Assigning a portion of an array disk to a virtual disk does not preclude the remaining portion of the array disk from being used by other virtual disks.) Only the virtual disks to which the hot spare is assigned will be rebuilt. When using Storage Management, a disk that is assigned as a hot spare on a PERC 3/Si, 3/Di, and CERC SATA1.5/6ch controller cannot also be used as an array disk.
|
Note When using the BIOS on a PERC 3/Si, 3/Di, or CERC SATA1.5/6ch controller, it may be possible to create a hot spare from an array disk that is also used in a virtual disk. To avoid confusion and maximize data protection, Storage Management does not allow an array disk to be both a hot spare and a member of a virtual disk. When assigning a hot spare, Storage Management displays the array disks that are not being used by a virtual disk. |
When assigning an array disk as a global hot spare on a PERC 3/Si, 3/Di, and CERC SATA1.5/6ch controller, the array disk should be as large or larger than the largest array disk on the controller.
You can assign the same dedicated hot spare to more than one virtual disk. In this case, the hot spare attempts to rebuild all portions of redundant virtual disks that reside on a failed array disk. To increase the likelihood that the hot spare is able to rebuild all virtual disks, you should do the following:
After the hot spare is activated to rebuild a particular virtual disk, it is no longer available for rebuilding other virtual disks should an additional array disk fail. For this reason, when a hot spare is activated it is automatically unassigned from the remaining virtual disks. To maintain data protection, you will need to add a new hot spare and assign it to the other virtual disks.
|
Note The "Assign and Unassign Dedicated Hot Spare" command is not available on the CERC SATA1.5/2s controller. |
Back to Storage Management Contents Page