Big Talk storage 4--raid disk array

RAID is an English redundant array of independent Disks (redundant array of independent disks), referred to as a disk array. Below are the various levels of RAID described below.

RAID0
Stripe (Stripe)Store. Theoretically, a RAID0 consisting of n disks is n times the read and write speed of a single disk. RAID 0 continues tobits or bytesSplit the data for the unit, read/write on multiple disks in parallel, so it has a high data transfer rate, but it does not have data redundancy, so it is not considered a real raid structure.


RAID1
Mirror Image (Mirror)Store. It is data redundancy through disk data mirroring, resulting in data that is backed up on paired independent disks. When raw data is busy, data can be read directly from the mirrored copy, so RAID 1 can improve read performance. RAID 1 is the highest unit cost in a disk array, but provides high data security and availability. When a disk fails, the system can automatically switch to read and write on the mirrored disk without having to reorganize the failed data.


RAID2
Hamming Code Check stripStore. The data is striped across different hard disks, and the Ribbon units arebits or bytes, using what is called a sea-plaintext to provideerror checking and recovery。 This coding technique requires multiple disk storage checks and recovery information, making RAID 2 technology more complex to implement and therefore rarely used in a business environment.


RAID3
parity (XOR) stripeStorage, shared check disk, data stripe storage unit bytes. It is very similar to Raid 2, where data is striped across different hard disks, except that RAID 3 uses simple parity and holds parity information with a single disk. If a disk fails, the parity disk and other data disks can regenerate the data, and if the parity disk fails, it does not affect data usage. RAID 3 provides a good transfer rate for large amounts of continuous data, but for random data, parity disks can be a bottleneck for write operations.



RAID4
parity (XOR) stripeStorage, shared check disk, data stripe storage unit is block. RAID 4 also tiles and distributes data across different disks, but the bars are in blocks or records. RAID 4 uses a single disk as the parity disk, and each write operation requires access to the parity disk, where the parity disk becomes the bottleneck of the write operation, so RAID 4 is rarely used in a commercial environment.



RAID5
parity (XOR) stripestorage, calibration data distributed storage, data stripe storage unit is block. RAID 5 does not specify a single parity disk, but instead accesses data and parity information across all disks. On RAID 5, the read/write pointer can operate against a list of devices at the same time, providing higher data traffic. RAID 5 is more suitable for small data blocks and random read and write data. The main difference between RAID 3 and RAID 5 is that RAID 3 involves all array disks for each data transfer, whereas for RAID 5, most data transfers operate on only one disk and can be performed in parallel. In RAID 5 There is a "write loss", that is, each write operation will produce four actual read/write operations, two reads the old data and parity information, two times write new data and parity information.
When recovering, for example, we need to recover the A0, it is necessary to need B0, C0, D0 plus 0 parity to calculate and derive A0 for data recovery. So when two discs are broken, the entire RAID data fails.

RAID6
parity (XOR) stripeStorage, two distributed storage checksum data, data stripe storage unit is block. RAID 6 adds a second independent parity information block compared to RAID 5. Two independent parity systems use different algorithms, and the data is very reliable, even if two disks fail at the same time without affecting the use of the data. However, RAID 6 requires more disk space allocated to parity information and a greater write loss than RAID 5, so "write performance" is very poor. Poor performance and complex implementations allow RAID 6 to be rarely used in practice.


RAID7
This is a new RAID standard, with its own intelligent real-time operating system and software tools for storage management, can be completely independent of the host running, do not occupy the host CPU resources. RAID 7 can be seen as a storage computer (Storage computer), which differs significantly from other RAID standards.

The raid 7 rating is by far the most theoretically performing RAID mode because it has been significantly different from the way it was formed. The basic form is shown in the past, a hard disk is an array of "pillars", and in RAID 7, multiple hard disks form a "pillar", they all have their own channels, and because of this, you can break this diagram into a hard drive connected to the main channel, just more than the previous level of subdivision. The advantage of this is that when reading/writing data for an area, you can quickly locate it, rather than just having access to a portion of the data area at the same time because of a single hard disk limit, in RAID 7, the previous single hard disk is the equivalent of splitting into multiple separate hard disks with its own read and write channels.



comparison of RAID10 and RAID01
-RAID10 is to do the mirror first, and then do the strip.
-RAID01 is to make a band first, and then do the mirror image.

For example, take 6 disks For example, RAID10 is the first to divide the disk into 3 groups of mirrors, and then to the 3 RAID1 do strip. RAID01 is the first use of 3 disk to do RAID0, and then the other 3 pieces of disk as a RAID0 mirror image. The following is a 4-disk example to describe the differences in security:

1, the situation of RAID10
In this case, we assume that when the DISK0 is damaged, in the remaining 3 disks, only when the DISK1 one disk fails, the entire RAID fails, we can simply calculate the failure rate of 1/3.

2, the situation of RAID01
In this case, we still assume that the DISK0 is corrupted and that the left stripe will not be readable. In the remaining 3 disks, as long as any one of the DISK2,DISK3 two disks is damaged, the entire raid will fail, and we can simply calculate the failure rate at 2/3.

So RAID10 is stronger than RAID01 in terms of safety.

From the logical location of the data store, RAID01 and RAID10 are identical under normal circumstances, and the number of IO generated by each read and write operation is the same, so there is no difference between read and write performance. When there is a disk failure, such as the previously assumed DISK0 damage, we can also find that in both cases, in the read performance will be different, RAID10 read performance will be better than RAID01.

Big Talk storage 4--raid disk array