Features at RAID levels

RAID, redundant arrays of inexpensive Disks, inexpensive redundant disk array. Also known as redundant arrays of independent Disks, an independent honor disk array.

The rationale is to use multiple smaller disks to form a large disk group in different organizational ways to improve the disk's IO capability and disk durability. Because of their different organizational ways, they are divided into different levels.

RAID has so far been divided into 0-6 seven levels.

standard RAID (raid-0,raid-1 ... RAID-6)

RAID-0

RAID0, also known as striped volume (strip) storage, is the basic principle of dividing data into segments or blocks of data that are stored on individual disks. The minimum number of disks used is 2, and the space available is n*min (disk 1,... Disk n). Min (Disk 1, ... Disk n) refers to the capacity of the smallest disk in n disks.

650) this.width=650; "title=" Raid0.png "alt=" wkiol1xd42bz0uzzaabh16zvsj0745.jpg "src=" http://s3.51cto.com/wyfs02/ M00/72/26/wkiol1xd42bz0uzzaabh16zvsj0745.jpg "/>

When data is written, the data is divided into segments of n (number of disks) and stored on the corresponding disk. Because the data is divided into n segments, its read and write speed is theoretically increased by N times. However, because this level does not provide data redundancy, it can cause data loss if a disk error occurs during the run.

RAID-1

RAID1, also known as mirrored volume (mirror) storage, is the principle that data is written to multiple disks equally, enabling data backup. The minimum number of disks used is 2, and the space available is 1*min (disk 1,... Disk n). Min (Disk 1, ... Disk n) refers to the capacity of the smallest disk in n disks.

650) this.width=650; "title=" Raid1.png "alt=" wkiom1xd4fjyir3faabkaps6tje208.jpg "src=" http://s3.51cto.com/wyfs02/ M00/72/2a/wkiom1xd4fjyir3faabkaps6tje208.jpg "/>

When the data is written, the data is copied to another N-1 disk, which realizes the data redundancy and improves the security of the data. But this way, the write speed is slow, the reading speed is fast, and the disk utilization is low.

RAID-5

RAID5 uses a checksum code and parallel transmission technology, the data and the corresponding XOR check code are stored on different disks, where all the XOR code occupies a disk space size, equivalent to the remaining N-1 block disk storage is data. This array requires a minimum of three disks.

650) this.width=650; "title=" Raid5.png "alt=" wkiol1xd6umxcoftaacowjzbqpe541.jpg "src=" http://s3.51cto.com/wyfs02/ M02/72/27/wkiol1xd6umxcoftaacowjzbqpe541.jpg "/>

Since RAID5 provides data validation, it does not affect the integrity of the entire data in the event of a disk problem, thus guaranteeing the security of the data. When a damaged disk is replaced, the data on the disk can be recovered by an XOR check code. However, the write rate is reduced because of the constant calculation of the relationship between the checksum code at the time of writing. For this array, you can use disk space of (N-1) *min (disk 1,... Disk n). Min (Disk 1, ... Disk n) refers to the capacity of the smallest disk in n disks. The minimum number of disks to use is 3 blocks.

RAID-6

raid6,independent data disks with two independent distributed parity schemes (stand-alone disk with two independent distributed calibration schemes). The RAID6 uses two independent checksum codes, and the data is very reliable, even if two disks fail at the same time, it will not affect the use of data and data integrity.

650) this.width=650; "title=" Raid6.png "alt=" wkiom1xd7ydiqxd8aade7q_jxmi656.jpg "src=" http://s3.51cto.com/wyfs02/ M00/72/2b/wkiom1xd7ydiqxd8aade7q_jxmi656.jpg "/>

With this array, at least 4 disks are required and the available data disk space is (N-2) *min (disk 1,... Disk n). Min (Disk 1, ... Disk n) refers to the capacity of the smallest disk in n disks. Of these, the disk space occupied by the two check codes is two disk space size. It reduces the write performance of the data, and its write performance is lower than RAID5. But its reading performance has improved.

Hybrid RAID

RAID-10

RAID10 As the name implies is to do RAID1 first, through the controller partition disk into a mirrored volume, for redundancy. Then do RAID0, through the controller to divide the data into strips for storage. To use a minimum of 4 disks.

650) this.width=650; "title=" Raid10.png "alt=" wkiol1xeclfwul-taac3x9drbjc139.jpg "src=" http://s3.51cto.com/wyfs02/ M01/72/29/wkiol1xeclfwul-taac3x9drbjc139.jpg "/>

With RAID10, this disk array improves the read and write performance of the data, but the space utilization of the disk is only 50% of the original. Redundancy, but only one copy of each data block and its data can be corrupted, and if damaged at the same time, the overall data will be corrupted. For example, when Disk 1 is damaged, the remaining three disks, only if Disk 2 is damaged, the entire data is lost, the theoretical failure rate of 1/3.

RAID-01

RAID10 As the name implies is the first to do RAID0, through the controller to divide the data into strips for storage. Then do RAID1, through the controller to partition the disk into a mirrored volume, for redundancy. To use a minimum of 4 disks.

650) this.width=650; "title=" Raid01.png "alt=" wkiom1xecntjgyskaac2peovkwi628.jpg "src=" http://s3.51cto.com/wyfs02/ M00/72/2c/wkiom1xecntjgyskaac2peovkwi628.jpg "/>

Using RAID01 this way, it also improves the data read and write rate and the data security, the disk space utilization is only 50% of the original. However, the theoretical failure rate has improved. For example, when Disk 1 is damaged, as long as Disk 2 or disk 3 is damaged, the entire data is lost, so its theoretical failure rate is 2/3.

This article is from "angry, bird" blog, please make sure to keep this source http://angrybird.blog.51cto.com/10503706/1688733

Features at RAID levels