Why Use RAID 6?
Solution
With the protection of RAID6, data can be reconstructed even if a second drive fails. With minimal performance impact and capacity consumption, RAID6 provides for extremely high data fault tolerance and can sustain multiple simultaneous drive failures.
RAID5 and RAID6 Comparison
RAID 5 RAID 6
Parity Single Dual
Protection One disk Two disks
Implementation Requires N+1 disks, minimum of 3 Requires N+2 disks, minimum of 4
Performance Medium impact on writes during rebuilds Greater impact on sequential writes than RAID5 during rebuilds
As described above, RAID5 protects against a single drive failure without downtime; however, if a second drive goes down, data will be lost. While two drive failures are less likely than one failure, the probability is an increasing concern based on the factors listed below:
Given the greater chance that two drives could fail simultaneously, it is clear to see the appeal of implementing RAID6 in a storage array.
Another factor to consider is that if an unrecoverable media error occurs on one of the members of a degraded array during a rebuild, that block has to be either declared 'lost' or the rebuild is aborted. This is due to the fact that there is not possible to regenerate the missing data with parity because the array is degraded. RAID 6 protects against this by having full parity available with a single disk failure. Of course, if the RAID 6 array is running in degraded mode with two failed disks, the potential for data loss is the same as a RAID 5, but the chances of a two-disk failure in 24 hours is very remote.
1. Growing adoption of less reliable drives: The benefits of SATA drives include lower prices and high capacity; however, their MTBF is lower than FC or SCSI drives. The increased use of these drives increases the probability of two drives failing at the same time.
2. High capacity means longer rebuilding time: Greater capacity on a single drive means more time is needed to rebuild data if one drive fails. Subsystems suffer heavy loading during the rebuild process and it's therefore more likely to damage another drive or for a second drive failure to occur during this longer rebuilding time.
3. Human error: When one drive fails, someone has to replace the damaged drive with a new one. An error, such as removing the wrong drive, can create other drive failures and data would be lost.
4. The failure rate significantly increases as the number of hard drives increases: Increasing the number of hard drives in an array effectively raises the expected failure rate of the first hard drive. During system recovery using a spare drive, the failure rate of the second hard drive is also multiplied. A subsystem composed of multiple hard drives needs added protection to guarantee data availability in case two drives fail simultaneously.
Given the greater chance that two drives could fail simultaneously, it is clear to see the appeal of implementing RAID6 in a storage array.
Another factor to consider is that if an unrecoverable media error occurs on one of the members of a degraded array during a rebuild, that block has to be either declared 'lost' or the rebuild is aborted. This is due to the fact that there is not possible to regenerate the missing data with parity because the array is degraded. RAID 6 protects against this by having full parity available with a single disk failure. Of course, if the RAID 6 array is running in degraded mode with two failed disks, the potential for data loss is the same as a RAID 5, but the chances of a two-disk failure in 24 hours is very remote.
