RAID controller failure is a cause of data loss. RAID arrays can be configured many different ways including:
- RAID 0: Stripe set, no parity. RAID 0 is optimal for speed but has no redundancy.
- RAID 10 or RAID 0 + 1: This utilizes the speed of RAID 0 and mirrors or clones the primary RAID 0 set to a secondary RAID 0 set.
- RAID 1: This is typically two drives. The primary drive is mirrored to the secondary.
- RAID 5: This is a stripe set with a distributed parity block which allows continued operation and the retention of data after the loss of one drive.
- RAID 6: This is a stripe set with dual distributed parity blocks which allows for continued operation and the retention of data after the loss of two drives.
The possibility of a hard drive failure should always be considered, however, RAID controller failure is a risk too; and it will result in the loss of accessibility to all data. Adaptec is the largest manufacturer of RAID controller cards. Adaptec RAID controllers are found in a variety of servers, SAN, NAS and other RAID devices from a host of OEM manufacturers including: Dell, HP and many others. Many Adaptec RAID controllers found in Dell, HP or other server manufacturers will host or present the RAID in different ways. For example HP RAID controllers use RAID 5 with delayed parity. Delayed parity means the parity block does not shift with each iteration of distributed reads, but is delayed. In contrast, an Adaptec RAID controller found in a Dell Power Edge (PERC) server does not use delayed parity and refers to the RAID 5 array as a container.
While Adaptec may be the largest, there are still many other manufacturers of RAID controllers including LSI, Intel, and Rocket RAID, just to name a few. All RAID controllers can present or host a RAID array in a variety of different ways. Some RAID controllers will house metadata on each drive in the array prior to the actual start of the array, while others will not. When a RAID controller fails there are several things to consider before recovering the data or attempting to restore the array.
- Was the array in a degraded state? Too often clients are not aware they were running with one drive out of the array when the second drive fails. This puts the user beyond RAID 5 redundancy thresholds and if the data is critical, it will need to be sent to a data recovery service. Prior to considering any action, view any and all logs and make sure the RAID 5 was not in a degraded state.
- What was the controller configuration? Some RAID controllers allow you to use only a portion of each hard drive in the array so the user can choose to make more than one array. For example a Dell Power Edge server could allow the user to make several containers using a portion of each drive for each container; therefore, the drives could host several unique RAID 5 arrays. You should be sure of how exactly the RAID controller was configured.
- Should you choose to try and re-introduce hard drives from a failed RAID controller into a new controller of the same type, it is very important to make images or clones of each drive. Should the introduction of hard drives from the failed RAID controller fail, data may have been written to the hard drives and compromised the integrity of the previous RAID configuration as well as the data house on disk.
As always, if the data is important, the drives from the RAID Controller Failure should be sent to a professional data recovery service.
In almost all cases where data is not recoverable from a RAID 5 controller failure, it was the user’s course of action after the controller failed that resulted in the destruction of the data.