Intel AFCSASRISER User Guide - Page 17

Hot Spare True fault tolerance requires the availability of a spare disk that the controller can add to the array and use to rebuild the array with the data from the failed drive. This spare disk is called a hot spare. It must be a part of the array before a disk failure occurs. A hot-spare drive is a physical drive that is maintained by the RAID controller but not actually used for data storage in the array unless another drive fails. Upon failure of one of the array's physical drives, the hot-spare drive is used to hold the recreated data and restore data redundancy. Hot-spare drives can be global (available to any array on a controller) or dedicated (only usable by one array). There can be more than one hot spare per array and the drive of the closest capacity is used. If both dedicated and global hot-spare drives are available, then the dedicated drive is used first. If the hot swap rebuild fails, then that hot spare is also marked failed. Since RAID 0 is not redundant, there is no hot spare value. If a hot-spare drive is not an option, then it is possible to perform a hot or cold swap of the failed drive to provide the new drive for rebuild after the drive failure. A swap is the manual substitution of a replacement drive in a disk subsystem. If a swap is performed while the system is running, it is a hot swap. A hot swap can only be performed if the backplane and enclosure support it. If the system does not support hot-swap drives, then the system must be powered down before the drive swap occurs. This is a cold swap. In all cases (hot spare, hot swap, or cold swap), the replacement drive must be at least as large as the drive it replaces. In all three cases, the failed drive is removed from the array. If using a hot spare, then the failed drive can remain in the system. When a hot spare is available and an automatic rebuild starts, the failed drive may be automatically removed from the array before the utilities detect the failure. Only the event logs show what happened. If the system is shut down during the rebuild, all rebuilds should automatically restart on reboot. Note: If running a sliced configuration (RAID 0, RAID 5, and RAID 6 on the same set of physical drives), then the rebuild of the spare will not occur until the RAID 0 array is deleted. On Intel® RAID Controller RS2WC080 and RS2WC040, if Virtual Drive is in degrade mode due to failed physical drive, auto rebuild is not supported for hot-plugged drive until a manual selection is made by users. As part of JBOD implementation for Intel® RAID Controller RS2WC080 and RS2WC040, all new drives that are hot-plugged will automatically become JBOD. Users need to manually move the JBOD drive to Unconfigured Good and auto rebuild starts after that. For more details, refer to Hardware User's Guide (HWUG) for above controllers. Data Redundancy Data redundancy is provided by mirroring or by disk striping with parity stripes. • Disk mirroring is found only in RAID 1 and 10. With mirroring, the same data simultaneously writes to two disks. If one disk fails, the contents of the other disk can be used to run the system and reconstruct the failed array. This provides 100% data redundancy but uses the most drive capacity, since 50% of the total capacity is available. Until a failure occurs, both mirrored disks contain the same data at all times. Either drive can act as the operational drive. • Parity is the ability to recreate data by using a mathematical calculation derived from multiple data sets. Parity is basically a checksum of all the data known as the Intel® RAID Software User's Guide 5