Home > Storage > Unity XT > Storage Admin > Dell Unity Dynamic Pools > Proactive copy and drive rebuilds
As mentioned previously, dynamic pools contain spare space extents to replace a drive going bad or a failed drive within the pool. The spare space extents can either be used as a target of a proactive copy operation, or a rebuild of a degraded RAID extent due to a faulted drive. In either case, spare space extents are used within the same drive partnership group as the bad drive.
The Dell Unity software tracks all drive errors, and if a drive is receiving errors above the systems internal thresholds, a proactive copy operation may be started. A proactive copy operation is the process of copying data from a drive going bad to a new location. Once the copy operation completes, the drive is failed by the software. Since this is a copy operation, the proactive copy completes much faster than a full drive rebuild. In the case of dynamic pools, a proactive copy operation copies a drive extent from the failing drive to a spare space extent within the same drive partnership group within the pool. During this operation, the system ensures the destination drive for each RAID extent does not already contain an extent from the same RAID extent. To ensure the proactive copy operation completes quickly, multiple drives extents can be copied from the failing drive at the same time. The target destination extents for proactive copy operations are also not selected from the same destination drive, which also spreads out the workload.
If a drive were to fail before it could be proactively copied, a rebuild operation must occur. Within a dynamic pool, a failed drive rebuild occurs by rebuilding the degraded RAID extents within the drive partnership group. A RAID extent is considered degraded when at least one of the drive extents within the RAID extent is no longer available. During the rebuild of a RAID extent, the remaining drive extents within the RAID extent are used to rebuild the missing drive extent to a spare space extent. Once complete, the spare space extent becomes part of the RAID extent and the RAID extent is no longer degraded. As the various RAID extents needing to be rebuilt and spare space extents are spread across many drives within the drive partnership group, many drives are engaged to complete the rebuild operations.
The number of drive failures that can be tolerated at the same time within a dynamic pool depends on the RAID protection and the location of the failures within the pool. As previously stated, the hot spare capacity setting does not improve the reliability of the RAID protection for the pool. For all RAID protections, if single drive failure occurs in different drive partnership groups, the rebuild operations can occur in parallel if enough spare space is present within the drive partnership groups the drives have failed in. For RAID 5, as long as two failures are not sustained within a single RAID extent, data access remains available. For RAID 6, two failures can occur within a single RAID extent without causing a data unavailable situation. Multiple drives within a single RAID 6 RAID extent can be rebuilt at the same time if enough spare space extents exist within the drive partnership group to rebuild both drives. For RAID 1/0, multiple failures can occur within the same RAID extent as long as a primary drive and its mirror are not lost within a single mirrored pair.
After the proactive copy or RAID extent rebuilds complete, the failed drive must be replaced in the system. At this time, the number of spare space extents has been reduced within the drive partnership group due to the failed drive and rebuild operations. If a free drive exists within the system, and it is the same size or larger and the same drive type as the failed drive, it will be consumed by the dynamic pool to replace the missing spare space extents. This operation closely resembles a single drive expansion, and the spare space extents will be rebalanced across the drives within the drive partnership group. Once the failed drive is replaced, it is left free within the system. If no free drives exist within the system, once the failed drive has been replaced, the new drive becomes consumed by the dynamic pool and the rebalance of spare space extents will occur.