When most people talk about the "enhanced reliability of RAID", what they
really are referring to is the fault tolerance of most RAID implementations.
Reliability and fault tolerance are not the same thing at all, as described in the discussion of RAID array reliability. Fault
tolerance refers to the ability of a RAID array to withstand the loss of some of its
hardware without the loss of data or availability.
When a fault occurs, the array enters a degraded state and
the failed drive must be replaced and rebuilt.
The capability of an array to tolerate hard disk faults depends entirely on the RAID level implemented. RAID 0, which has no redundant
information, has no fault tolerance for that reason; if any drive fails the array goes
down. RAID levels 1, 2, 3, 4, 5 and 7 can tolerate the loss of one hard disk; RAID 6 can
tolerate the loss of two. The multiple RAID levels can
often tolerate the loss of multiple hard disks depending on which ones they are and how
the array is configured. For example, if you have a RAID 10 array consisting of two pairs
of mirrored drives striped together, the array can tolerate the simultaneous loss of two
of the four drives, as long as they are not in the same pair.
Ultimately, fault tolerance of a RAID-equipped PC depends not only on the reliability
of the drives, but also the other hardware in the system. For example, most RAID
implementations are dependent upon the RAID controller not failing; if it goes down then
your data is probably fine, but availability is not--this is one reason why duplexing is sometimes used, since it can tolerate the loss
of a drive and a controller. Since items such as power supplies have a high
failure rate and ultimately can bring down any array, fault tolerance in a PC equipped
with RAID is also often used to describe features implemented on the system as a whole to
prevent faults in this support equipment; see here
for more details.
Next: Availability
