RAID Levels features

RAID is an acronym for redundant arrays of inexpensive disks.

Early in order to provide inexpensive solutions to improve the i/0 capacity and durability of the disk, in the 90 era of raid occurred. The principle of RAID technology is to use multiple disks as a piece of hard disk, to improve I/O capability by parallel operation, and to provide redundancy to improve durability. Depending on the organization, the raid type is 0-6,raid10,raid01,raid50,raid7, and the different RAID types apply to different scenarios.

Raid0
RAID0 is also known as strip, which means striped coils. When the data is stored, the RAID0 controller will store contiguous data blocks, and the data blocks are stored separately on the RAID0 disks. RAID0 's ability to read and write is much better than a single disk, and it is also the best disk parallel read and write performance in raid. But it has a fatal flaw, when a hard drive is damaged, all the data is lost, so the data is very unreliable. RAID0 requires at least two hard disks, and the disk group has a free space of min (s1,s2. SN) *n. The comparison applies to scenarios where the security of the data is not required.

Raid1
RAID1 and RAID0 are the two extremes in raid, RAID1 provides full data backup, and the data to be stored is stored in multiple copies (usually RAID1 only two hard drives), and each hard disk stores one copy of the data. RAID1 provides data redundancy through mirroring, and disk space utilization is not high, only min (s1,s2. SN). In addition, for RAID1, the read performance is improved because it can read data in parallel, but the write performance is slightly lower than that of a single disk. More suitable for storing critical data, many hard drives that are mirrored are less likely to be broken.

raid2,3 are not commonly used, they are all in the RAID0 and RAID1 to seek a balance point.

Raid4
Due to the low space utilization of the RAID1, a single hard disk is used to make the check disk in Raid4 to realize redundancy. Take three disk (A,B,C) as an example, when the data is stored, the data is divided into 2 blocks (A, b), the controller to A A, B to do XOR operation, get C, finally let A deposit A, B and save B,c c. RAID4 data is sliced and accessed concurrently, and read and write performance is improved. However, because the checksum data exists on a disk, and the data read and write need to verify the data, the check disk will be unusually busy, may become a performance improvement bottleneck. The RAID4 provides fault tolerance, allowing only one hard drive to be damaged. When the hard disk corruption, RAID4 will be degraded work, the hard work pressure will become larger, it is recommended to replace the hard disk immediately or when there is a hard disk damage, the entire disk group data will not be used. Raid4 need to have at least 3 pieces of disk, the extra 3 is also OK, the extra hard disk can do spare disk, do backup use. The available space for Raid4 is (n-1) *min (S1,S2). SN).

Raid5
RAID5 and Raid4 very similar, the only difference is not to take out the hard disk to do the check disk, instead, the data and the verification data stored together, verify that the data stored in rotation on each disk. RAID5 more than Raid4, there is no performance bottleneck of the calibration disk, because the reading pressure of the check data is divided by each hard disk, the overall stability has improved.

Raid6
The RAID6 is improved on the RAID5, and the checksum is stored two times.

Raid10
RAID10 's work is to first group the hard disk 22, the group of hard disks to mirror each other, the hard disk between the groups in accordance with the RAID0 working mode. RAID10 must have at least 4 hard drives, only one damage per set of hard drives, free space on disk for min (s1,s2,.. SN) *N/2, improved read and write performance.

Raid01
RAID01 is also the hard disk 22 group, but its group to do the sub-volume, between the group to do mirror redundancy. Its fault tolerance is that when a packet within a disk corruption (regardless of how many), the data is still safe, and when multiple groups have disk corruption at the same time, the data security is relatively suspended.

Raid50
RAID50 first the hard disk every three divided into a group, each group of hard disk in accordance with RAID5 mode of work, the hard disk between the group to work according to RAID0 mode. RAID50 's data redundancy capability is not strong, but its read-write
Performance is improved. Suitable for scenarios where there is no requirement for data security, but high requirements for I/O performance.

This article is from the "Looking at the Stars" blog, please be sure to keep this source http://nick121.blog.51cto.com/10716426/1695507

RAID Levels features