Appendix C RAID
12
Appendix C RAID
C
C
C.
.
.1
1
1
R
R
RA
A
AI
I
ID
D
D
The word RAID was first introduced in the thesis “A Case for Redundant Arrays of Inexpensive Disks”
released by professors of the University of California at Berkeley of the U.S.A. in 1987.
The initials RAID of “Redundant Arrays of Inexpensive Disks” mean literally “a set of redundant and
inexpensive disks”. However, RAID has been generally explained with “Inexpensive” replaced by
“Independent” in these days. Large-capacity hard disks were very expensive at that time, but I/O
performance was not high enough for the costs. Disk access was a bottleneck in system performance.
RAID was created for the following purposes:
y Reducing costs by using inexpensive small-capacity disks
y Improving performance by concurrently accessing multiple disks
y Increasing reliability by adding redundant data
RAID levels 1 to 5 are defined in the thesis mentioned above. However, the table below shows the
features of RAID0 (striping), RAID1, RAID5, RAID10, RAID50, and RAID6 only. RAID0 (striping)
has no redundancy and does not satisfy all requirements for RAID, but it is in wide use. RAID10 is a
combination of RAID0 and RAID1. RAID50 is a combination of RAID0 and RAID5. RAID6 can
keep redundancy by its double-parity feature even if one physical disk becomes faulty.
Each RAID number is used simply to sort out data division methods or repair methods, and the RAID
numbers do not indicate any priority.