There are several benefits to using a RAID (Redundant Array of Independent Disks) such as the ability to create large data volumes, higher data transfer speeds and fault tolerance. Each RAID configuration is a different level of compromise between data security, data transfer bandwidth and the amount of hardware required.
Any budget constraint will usually determine the best RAID system especially if you have a particular requirement for the setup. When money is no object however, fault tolerance should almost certainly be one of the highest priorities. Although a RAID array provides fault tolerance you should always have a robust backup strategy in place, to ensure that if your RAID does fail, it can be recovered without sending the array for data recovery.
100% Redundancy from RAID 10
A RAID 10 array combines the mirroring of RAID 1 with the data striping of RAID 0, but does have the drawback of requiring a pair of disks for each stripe using in the configuration. This provides the array with full 100% redundancy, giving it the highest possible level of fault tolerance. If any drive within a mirrored pair of drives fails, the RAID will continue to operate, allowing the faulty drive to be replaced with the data re-mirrored to restore the full redundancy. When one drive in a pair fails, there is high chance of the other drive also failing before the rebuild can be completed.
As there is no need to recalculate a parity stripe when writing data to a RAID 10 drive, with mirroring automatically taken care of by the RAID unit, the read and write speeds achievable may be higher. Providing a mirrored pair of drives does not fail, it is possible for a RAID 10 array to remain operational with up to half of the drives not working. Running a RAID 10 array in this degraded would put your data at serious risk. Exchange and SQL databases hosted on high availability servers are the most common uses for a RAID 10 array, although they are becoming an increasingly popular setup for web servers.
Higher Capacity from RAID 5
The RAID 5 configuration stripes the data across a set of drives using one of the drives within each data slice to store the parity information, which rotates across the set of drives. In the event of a single drive failing the data stored on the failed drive can be reconstructed using the parity information, which allows the RAID array to operate in degraded mode. By only using a single drive to store the parity data, larger data volumes can be created for any given set number of hard disk drives.
Although a RAID 5 array is able to operate in degraded mode when a single drive fails, it should be replaced as soon as possible with the data rebuilt to it. Running in degraded mode also causes a performance hit, as well as the imminent risk of total failure of the array should another drive fail. If another drive fails it will result in the RAID going offline. Despite the increased drive capacities which have resulted in long rebuild times, the use of RAID 5 is still extremely common, as it provides a good compromise between capacity, data transfer speed and data security.
RAID Array Data Recovery
Even though RAID 10 arrays are extremely robust through the use of mirrored pair they are still occasionally seen for data recovery. Unless the failures have been ignored, resulting in one drive in a mirrored pair holding out-of-date data, the use of RAID 10 doubles the chance of successfully recovering all the data from each slice of the RAID. Our data recovery success rate for RAID 10 is extremely high.
Despite there being a higher level of risk attached to using a RAID 5 array, the data recovery success rate is also very high. Although we see many more RAID 5 arrived in the laboratory for data recovery, it is unusual for the drive failures to result in the loss of large areas of raw data sectors.
It is always better to use fault tolerance than none at all. Which choice is best is usually driven by your budget and the level of risk you feel is appropriate to take. The financial implications of any data loss on your company, even if temporary, should be considered. A backup strategy should also be put in place to ensure the highest possible chance of avoiding any data loss.