AHCI and RAID on HP Compaq Elite 8000, 8100, 8200, and 8300 Business PCs - Technical White Paper

11
RAID 5 with three hard drives
RAID 5 has been used in servers for many years and is one of the most common types of RAID. RAID 5 uses striping with
parity data in distributed blocks across all member disks. Therefore, the mass storage controller can simultaneously
write new information to two hard drives and parity information to the third hard drive, so if one hard drive fails, the
RAID controller can rebuild all the information after the volume degradation occurred. Hence, RAID 5 with three hard
drives has similar performance to RAID 0 with two hard drives, and the reliability of RAID 1 with a minimum of three hard
drives.
Table 5: RAID 5 with three hard drives (parity)
First disk Second disk Third disk
Data Segment 1 Data Segment 2 Parity for 1 and 2
Data Segment 3 Parity for 3 and 4 Data Segment 4
Parity for 5 and 6 Data Segment 5 Data Segment 6
Data Segment 7 Data Segment 8 Parity for 7 and 8
Data Segment 9 Parity for 9 and 10 Data Segment 10
Parity for 11 and 12 Data Segment 11 Data Segment 12
Data Segment 13 Data Segment 14 Parity for 13 and 14
Data Segment 15 Parity for 15 and 16 Data Segment 16
Parity for 17 and 18 Data Segment 17 Data Segment 18
In the previous table, each “Data Segment x” represents a strip. A stripe is made with strip “Data Segment x”, “Data
Segment y” and strip “Parity for x and y.” Notice that the strip “Parity for x and y” is used to store the required
information to recreate the data if any one of the RAID volume members is compromised.
To better illustrate the concept of RAID 5 (parity-based RAID), Figure 4 shows how a sequence of data “ABCD...” is stored
in a RAID 5. In this example, each letter represents a segment or strip. The various pieces of the information go to
different hard drives, hence if any one member of the RAID 5 fails, the information from all members is not lost.
Figure 4: RAID 5 with three hard drives