Databases Reference
In-Depth Information
RAID dei nes two ways to provide failure protection: disk mirroring and parity generation. RAID
1, often called
disk mirroring
, places data in equal parts on separate physical disks. If one disk fails,
the array controller will mirror data from the remaining good disk onto a new replacement disk.
Figure 4-5 details the frequent combination of mirroring and striping. This system is often called
RAID 1 + 0 or simply RAID 10.
Data 1
Data 1
Data 2
Data 2
Data 3
Data 3
Data 4
Data 4
Data 5
Data 5
Data 6
Data 6
Data 7
Data 7
Data 8
Data 8
FIGURE 4-5
The storage array uses an
exclusive or
(XOR)
mathematical calculation to generate parity data.
Parity enables the array to recreate missing data by combining parity information with the data that
is distributed across the remaining disks. This parity data enables you to make efi cient use of your
capacity but at the cost of performance, as the XOR calculation needed to generate the parity data is
resource intensive. More details about the performance implications of data protection can be found
in the Disk Drive Performance section of this chapter.
Many different parity RAID coni gurations have been dei ned. The two most common types are
disk striping with parity (RAID 5) and disk striping with double parity (RAID 6). Examples of both
are shown in Figure 4-6 and Figure 4-7. RAID 5 protects a system against a single disk drive failure.
RAID 6 protects against a double disk failure. RAID 5 and 6 offer disk failure protection while
minimizing the amount of capacity dedicated to protection. Contrast this with RAID 1, which con-
sumes half of the available storage in order to protect the data set.
Data 1
Data 2
Data 3
Parity 1
Data 4
Data 5
Parity 2
Data 6
Data 7
Parity 3
Data 8
Data 9
Parity 4
Data 10
Data 11
Data 12
FIGURE 4-6
Search WWH ::
Custom Search