RAID (Redundant Array of Independent Disk) refers to a technology in computer data storage used to implement the fault tolerance feature of computer storage media (especially hard disk) by means of data redundancy, either by using software, As well as a separate RAID hardware unit. RAID also has several abbreviations of Redundant Array of Inexpensive Disks, Redundant Arrays of Independent Drives, as well as Redundant Arrays of Inexpensive Drives. This technology divides or replicates data into multiple separate hard drives. RAID is designed to improve data reliability and improve I/O performance of hard disks.
Disk is at risk for damage. This damage can result in performance loss or loss of data. Although there are data backups, there is still the possibility of missing data due to changes after the last time the data is backed up. Therefore the reliability of a disk must be continuously improved. Various ways are done to improve performance and also the reliability of the disk. Usually to improve performance, many disks are involved as one storage unit. Each data block is split into several sub-blocks, and subdivided into those disks. When sending data these disks work in parallel, so it can increase the transfer speed in reading or writing data. Coupled with the design on the rotation of each disk, the performance of the disk can be improved. This method is known as RAID. It can also increase the reliability of the disk by performing data redundancy.
Three properties of RAID :
- RAID is a set of disk drives that are considered to be a single disk system.
- The data is distributed to the physical drive array.
- Redundant disk capacity is used to store parity information, which ensures data recovery when a problem or disk failure occurs.
So, RAID is one of the answers to the problem of memory disk gap speeds with the CPU by replacing large-capacity disks with small disks and distributing data on those disks in such a way that they can be read back later.
RAID can be divided into 8 different levels Level 0, level 1, level 2, level 3, level 4, level 5, level 6, level 0 + 1 and 1 + 0. Each level has its advantages and disadvantages.
1. RAID 0
RAID level 0 uses a collection of disks with striping at block level, without redundancy. So just keep doing a blocking of data blocks into multiple disks. This level is not actually included in the RAID group because it does not use redundancy for its performance improvement.
2. RAID 1
This level 1 RAID is a mirroring disk, duplicating each disk. This can improve disk performance, the amount of disk needed to be doubled, so the cost becomes very expensive. At level 1 (disk duplexing and disk mirroring) on the hard disk partitions are copied to a partition on another hard disk because if one is damaged, it is still available in the mirror.
3. RAID 2
This level 2 RAID is an organization with error-correcting-code (ECC). As in the memory where the detection by error using parity bits. Each data byte has a corresponding bit parity that represents the number of bits in the data byte in which the parity bit = 0 if the number of even bits or parity = 1 if odd. Thus, if one of the bits in the data is changed, the parity changes and does not match the parity of the stored bit. Thus, an error occurs on a single disk error, the data can be reshaped by reading error-correction bits on another disk.
4. RAID 3
Level 3 RAID is an organizing with bit interleaved parity. This organization is almost the same as RAID level 2, the difference is that this level 3 RAID only requires a redundant disk, regardless of the number of disks. So do not use ECC, just use a little parity for a set of bits that have the same position on each disk that contains data. It also uses data striping and diskette access in parallel.
5. RAID 4
RAID level 4 is an organizing with blocked interleaved parity, which uses data striping at the block level, storing a parity block on a separate disk for each data block on the corresponding disks. If a disk fails, the parity block can be used to reshape the data blocks on the failed disk. The transfer speed for reading data is high, because each data disk can be accessed in parallel. Likewise with writing, because the data disk and parity can be written in parallel.
6. RAID 5
RAID level 5 is an organizing with blocked interleaved blocked parity. Data and parity are distributed on all disks including an additional disk. On each block, one of the disks stores parity and the other disk stores data. For example, if there is a collection of 5 disks, the parity of blocks to n will be stored on disk (n mod 5) + 1; The nth block of the other four disks stores the actual data of the block. A block parity can not be used for data on the same disk, since a disk failure will cause the data to be lost along with its parity and the data can not be assigned. The parity distribution on each disk avoids the use of a disk parity as at RAID level 4.
7. RAID 6
RAID level 6 is also called P + Q redundancy, such as RAID level 5, but it stores additional redundant information to anticipate the failure of multiple disks at once. RAID level 6 performs two different parity calculations, then stored in separate blocks on different disks. So, if the data disk is used as much as n disk pieces, then the amount of disk needed for this level 6 RAID is n + 2 disk. The advantage of this level 6 RAID is the extremely high data reliability, because to cause data to be lost, failure must occur on three disks in the mean time to repair interval (MTTR). The disadvantage is the time penalty at the time of writing the data, because each writing is done will affect the two parity blocks.
8. RAID level 0+1 dan 1+0
RAID level 0 + 1 and 1 + 0 is a combination of RAID levels 0 and 1. RAID level 0 has good performance, while RAID level 1 has reliability. However, in the second test it is the same thing. In RAID 0 + 1, the disk assemblies are stripped, and the disk strip is mirrored to other disks, producing the same data strips
The other combination is RAID 1 + 0, where the disks are mirrored in pairs, and then the result of the mirrored pair is stripped. RAID 1 + 0 has an advantage over RAID 0 + 1. For example, if a disk fails on RAID 0 + 1, the entire strip is inaccessible, only a portion of the strip is accessible, whereas in RAID 1 + 0, the failed disk is inaccessible, but the mirror pair is still Accessible, that is, disks other than disks that fail.