Databases Reference
In-Depth Information
Figure 13.10
RAID 5.
RAID 6 (Figure 13.11) extends the ideas of RAID 5 by adding a second parity block
within each data stripe on the storage array. The introduction of a second parity block for
each stripe set allows RAID 6 to be particularly resilient to disk failure, being able to sus-
tain two disk failures within the array and still operate normally. RAID 6 is not particu-
larly efficient with a small number of disks because too high a percentage of the blocks
are used for parity, increasing cost and decreasing performance. However, for wide arrays
the overhead becomes more reasonable, and the benefits can become significant enough
to make this a winning strategy. This is particularly important because as the stripe set
becomes wider, the MTTF decreases, as discussed above.
13.7.7 RAID 1+0
RAID level can be used in combination and the main combinations are the combina-
tion of RAID 0 (striping) with either RAID 1 or RAID 5. Formally, the combination of
RAID 1 and RAID 0 is denoted RAID 1+0, though the short form of RAID 10 has
become popular. In RAID 10 data is mirrored and striped as shown in Figure 13.12.
13.7.8 RAID 0+1
A variation of RAID 10 is RAID 0+1, where data is striped and the striping sets are mir-
rored as shown in Figure 13.13.
