Dell POWERVAULT MD3600I - Manual (Page 24)

NOTE: It is recommended that you run data consistency checks on a redundant array at least once a month. This
allows detection and automatic replacement of unreadable sectors. Finding an unreadable sector during a rebuild
of a failed physical disk is a serious problem, because the system does not have the redundancy to recover the
data.
Media Verification
Another background task performed by the storage array is media verification of all configured physical disks in a disk
group. The storage array uses the Read operation to perform verification on the space configured in virtual disks and the
space reserved for the metadata.
Cycle Time
The media verification operation runs only on selected disk groups, independent of other disk groups. Cycle time is the
time taken to complete verification of the metadata region of the disk group and all virtual disks in the disk group for
which media verification is configured. The next cycle for a disk group starts automatically when the current cycle
completes. You can set the cycle time for a media verification operation between 1 and 30 days. The storage controller
throttles the media verification I/O accesses to disks based on the cycle time.
The storage array tracks the cycle for each disk group independent of other disk groups on the controller and creates a
checkpoint. If the media verification operation on a disk group is preempted or blocked by another operation on the disk
group, the storage array resumes after the current cycle. If the media verification process on a disk group is stopped
due to a RAID controller module restart, the storage array resumes the process from the last checkpoint.
Virtual Disk Operations Limit
The maximum number of active, concurrent virtual disk processes per RAID controller module installed in the storage
array is four. This limit is applied to the following virtual disk processes:
• Background initialization
• Foreground initialization
• Consistency check
• Rebuild
• Copy back
If a redundant RAID controller module fails with existing virtual disk processes, the processes on the failed controller
are transferred to the peer controller. A transferred process is placed in a suspended state if there are four active
processes on the peer controller. The suspended processes are resumed on the peer controller when the number of
active processes falls below four.
Disk Group Operations
RAID Level Migration
You can migrate from one RAID level to another depending on your requirements. For example, fault-tolerant
characteristics can be added to a stripe set (RAID 0) by converting it to a RAID 5 set. The MD Storage Manager provides
information about RAID attributes to assist you in selecting the appropriate RAID level. You can perform a RAID level
migration while the system is still running and without rebooting, which maintains data availability.
24