Redundant array of independent hard disks (RAID, Redundant Array of Independent Disks), old name Redundant array of Inexpensive disks (Redundant Array of Inexpensive Disks), referred to as a disk array . The basic idea is to combine a number of relatively inexpensive hard drives into a hard disk array to achieve even more expensive, large-capacity drives. Depending on the version selected, RAID has the benefit of one or more of the following aspects than a single hard drive: Enhanced data integration, enhanced fault tolerance, increased throughput or capacity. In addition, the disk array looks like a separate hard disk or logical storage unit for a computer. Divided into raid-0,raid-1,raid-1e,raid-5,raid-6,raid-7,raid-10,raid-50,raid-60. (The above is from wiki Chinese)
RAID0
All drives in parallel, read and write parallel processing, with the fastest read and write speeds in all raid. Any piece of disk offline, all data can not be restored.
RAID1
All hard drives mirror each other, theoretically reading speed is linearly related to the number of hard drives, and the write speed is equal to the speed of a single hard drive. As long as one hard drive is online, it will work properly. Highest security in all raid. The more hard drives, the more wasted storage space, the least space utilization in all of the raid.
Raid2/raid3/raid4
Less practical applications, mainly used in research areas. Similar in principle to RAID5, performance is inferior to RAID5.
RAID5
RAID5 is a storage solution that combines storage performance, data security, and storage costs. Each hard drive has a parity zone and requires a minimum of three hard drives. For drives A, B, C in the array, each drive is divided into three zones: A1, A2, A3, B1, B2, B3, C1, C2, C3, where A3, B2, C1 are parity zones, A3=B3⊕C3,B2=A2⊕C2,C1=A1⊕B1, according to the characteristics of XOR or operation, if a⊕b= C is a=b⊕c and B=a⊕c, when the hard disk B fault, just calculate b3=a3⊕c3, B1=A1⊕C1,B2=A2⊕C2 can complete data recovery. It can be understood that when a hard disk fails, it is necessary to solve the data recovery by solving n unary one-time xor or equation.
|
A |
B |
C |
1 |
A1 (data) |
B1 (data) |
C1 (A1⊕B1) |
2 |
A2 (data) |
B2 (A2⊕C2) |
C2 (data) |
3 |
A3 (B3⊕C3) |
B3 (data) |
C3 (data) |
RAID6
Similar to RAID5, but each hard drive has two separate check zones, requiring a minimum of four drives. Two check area pq,p and RAID5 similar, do the XOR operation, Q need to apply field theory (Galois filed gamma MabThera domain, try to query this information, but require mathematical expertise, I have limited knowledge of mathematics, the document read very difficult, no understanding). Similar to RAID5, it can be understood that when there are two hard disk failures, it is necessary to solve the data recovery of n two yuan once the different or equation group.
RAID10
Do a mirror and then partition, as long as each partition has at least one hard drive online can work, high reliability.
RAID01
First do the partition and then do the mirror, the equivalent of multiple RAID0 array mirror, as long as a RAID0 in a hard disk failure, the entire RAID0 will fail, relative to the RAID10 reliability is low.
RAID50
Do RAID5 then do RAID0, each RAID5 partition allows at least one hard drive failure, when there is a RAID5 partition with at least two hard disk failures, the entire RAID50 will be invalidated.
RAID60
Do RAID6 then do RAID0, each RAID6 partition allows at least two hard disk failures, when there is a RAID6 partition with at least three hard drive failure, the entire RAID60 will be invalidated.
RAID level |
Minimum hard drive |
Space efficiency |
Maximum fault tolerance |
Array failure efficiency |
Read performance |
Write performance |
Security |
Objective |
0 |
2 |
1 |
0 |
(1-R) n |
N |
N |
No |
The pursuit of maximum capacity, the fastest speed |
1 |
2 |
1/n |
N-1 |
Rn |
N |
1 |
highest, at least 1 block hard Drive online |
The pursuit of the highest data security |
1+0 |
4 |
N/2 |
N/2 |
|
N |
N/2 |
each group at least 1 block hard disk online |
combines the benefits of a with a waste of storage |
5 |
3 |
1-1/n |
1 |
1-nr (1-r) n-1-(1-r) n |
n-1 |
n-1 |
Allow 1 block hard drive offline |
pursue maximum capacity minimum budget |
6 |
4 |
1-2/n |
2 |
1-n (n-1) R2 (1 -R) n-2-(1-r) n |
n-2 |
N-2 |
Allow 2 block hard drive offline |
Same as RAID5 , more secure |
5+0 |
6 |
(1-1/n)/2 |
|
|
n-m |
n/m-1 |
each group up to 1 block hard disk offline |
performance is higher than RAID5 , but each packet is wasted 1 block hard disk |
6+0 |
8 |
(1-2/n)/2 |
|
|
N-2m |
N/m-2 |
Up to 2 per chunk block hard Drive offline |
with 5+0 , security is higher, but each packet is wasted 2 Block Hard disk |
Recommendations for the immature selection of raid:
If the storage space is sufficient, select Raid1+0
A separate database server its disk array is best chosen raid1+0
If you choose Raid1+0 Not enough space, choose RAID 5
If the disk array contains more than 8 disks, it is best to select RAID 6
Finally, attach the link to the wiki on the raid:
English Https://en.wikipedia.org/wiki/Standard_RAID_levels
Chinese Https://zh.wikipedia.org/wiki/RAID
About RAID Summary