Detailed RAID technology

 Source: Your blog network (yourblog.org)
RAID technology Detailed full name
RAID Redundant Array of Disks, is a ;) In 1987 abbreviations by the Patterson, Gibson and Katz at the University of California, Berkeley branch of the definition of an article. RAID disk array technology allows a group to achieve the necessary data protection and data redundancy, and formed to improve the performance of data read and write band distribution. RAID was originally used for high-end server market, but with the rapid development of computer technology, RAID technology has penetrated into all areas of the computer over. Now, in the home computer motherboard, RAID controller chip can be seen everywhere.
general, RAID systems can exist in a variety of Interface to us for the time being, PATA, SATA and SCSI hard disk can have the appropriate composition of RAID. With the Intel 865/875 series chipsets Group release, domestic market are turning to hard disk interface, SATA, PATA and RAID mode will also benefit from the transition to SATA.
RAID technology with people's use of the process, through a series of changes and development. and in the home market , we generally see only RAID 0, RAID 1 and RAID 0 +1 these types of disk array mode. but from the birth of DFI Lanparty board, we have ushered in the fourth disk array mode, that is, RAID 1.5.
from the actual application, the household formation of the majority of RAID is to further improve the performance of disk read and write, and data backup to reach by other means (such as burning). So, there are only two hard cases, people are willing to try to RAID 0-based, but the birth of RAID 1.5 we changed the concept. What these two relatively inexpensive way of disk array performance with what? Let us as we announced.
RAID 0 :
RAID 0 uses a called a block), the data is divided into from 512 bytes to several megabytes of a number of pieces, and then alternately written to disk. The first one is written to disk 1, the first two were written to disk 2, and so on . When the system reaches the end of a disk array, it is written to disk 1, the next section, it goes on.
segmentation data can be I / O load evenly to all drives. Since the drive can write or read, making the performance improved significantly. However, it has no data protection. If one disk fails, then the data will be completely lost. Therefore, RAID 0 does not apply to mission-critical environment, but it is very suitable for video, image production and editing.
RAID 1:
RAID 1 also known as the mirror, because the data on a disk is completely copied to another disk. If a data disk error, or hard disk there have bad, then another hard disk failure can be remedied due to return to data loss and system outages. In addition, RAID 1 also allows duplex - that you can copy the entire controller, so a disk failure or controller failure , your data can be protected. mirror and duplex disadvantage is the need to double the number of multi-drive to copy data, but the system does not read and write performance and the resulting increase, this may be an amount not small expenditure. RAID l can be achieved by software or hardware.
RAID 2:
RAID 2 is the development of mainframes and supercomputers with Hamming code parity disk array. disk drive group, the first one , No. 2, No. 4 hh n-2 disk drive power of a dedicated parity disk is used for calibration and error correction. the following diagram: the formation of the seven disk drives RAID 2, 1,2, 4 disk drives (red) is the error correction disk, the rest of the (purple) used for storing data. RAID 2 read and write large amount of data has a very high performance, but a small amount of data read and write performance but not good, so RAID 2 uses less real.
the particularity of the RAID 2, as long as we use more disk drives, checking account in which the percentage of disk less. If you want to achieve the ideal speed and better disk utilization rate, it is best to increase the ECC code stored checksum hard, but it will cost to pay more to buy the hard drive to ensure data redundancy. For the controller design, it said than the following RAID 3 , 4 or 5 should be simple.
RAID 3:
RAID 3, or with dedicated parity bit (parity) bands. Every chip has a band equivalent to a space used to store redundant information, that is, a parity bit. parity bit is encoded message data if a disk error, or a disk fails, you can use it to restore the data. In the data-intensive environment or a single user environment , the formation of RAID 3 long continuous record of the visit is beneficial, but as with RAID 2, visit the shorter records, performance will decline.
RAID 4:
RAID 4 is the parity-check codes with independent disk structure. It is similar to RAID 3, RAID 4 is different access to the data is carried out according to the data block. RAID 3 is a one bar, and a vertical bar a RAID 4. So often have access to RAID 3 array All the hard drives, and RAID 4 only access useful hard drive. so read data at a rate greatly increased, but in the written data is required from the data in the hard disk drive and check the hard drive to recover the old data and new data validation, and then test the updated bit write data and hard drives, so the processing time is longer than the RAID 3.
RAID 5:
RAID 5 is also called with distributed parity bands. Each section has put the equivalent of a disk, greatly reducing the burden of parity disk. Despite the loss of some capacity, RAID 5 was able to provide the overall performance of the more perfect, which is widely used as a disk array solutions. It is suitable for input / output intensive, high read / write ratio of applications such as transaction processing.
level in order to have RAID 5 redundancy, we need at least three components of the disk array disk. RAID 5 disk array controller hardware can be achieved by , or by some of the network operating system software.
RAID 6:
RAID 6 is the distribution of storage with two independent parity disk structure code. It uses the distribution of the disks in different The second parity to achieve enhanced RAID 5. It can withstand multiple simultaneous drive failures, but the value used to calculate the parity data and verify the correctness of the time it takes more, resulting in a system load heavier, significantly reduce the overall disk performance, and that requires a very complex system controller. Of course, the introduction of a second parity value, we therefore need is N +2 disks.
RAID 7:
RAID 7 itself with intelligent real-time operating systems and software for storage management tool that can run completely independent of the host, do not take up host CPU resources. RAID 7 storage computer operating system (Storage Computer Operating System) is a Real-time event-driven operating system, mainly used for system initialization and RAID 7 disk array arrangements of all data transmission, and convert them to the appropriate physical storage drive. by Storage Computer Operating System to set and control the access speed enable the host I / O transfer performance to the best. If one disk fails, recovery operations can be performed automatically, and manage the reconstruction process of the backup disk.
RAID 7 use of a non-synchronized access mode, greatly reducing write data bottlenecks and improve the I / O speed. (so-called non-synchronous access, that is, RAID 7 for each I / O interface has a dedicated high-speed access, as the data path or control the flow of information, it can be controlled independently own systems of data for each disk access.) If there are N disks RAID 7, then remove a check disk (for redundancy terms), but can handle N-1 hosts were issued by the system read / write instructions, thereby significantly improving the I / O applications. RAID 7 systems built-in real-time operating system can also automatically sent from the host read / write commands to optimize the processing to intelligently read data may be read into the pre- fast cache, thus greatly reducing the number of head rotation and improve the I / O speed. RAID 7 to help users effectively manage the increasingly large data storage systems, and to improve the efficiency of the system at least more than double to meet the class the different needs of users.
RAID 10 (RAID 0 +1):
RAID 10, also known as the mirror array bands, and now we generally call it RAID 0 +1. RAID 10 (RAID 0 +1) to provide 100% data redundancy and support greater volume size. build RAID 10 (RAID 0 +1) requires 4 disks, two of them into strips and data distribution, providing read and write performance of RAID 0, while the other two was a mirror in front of the two hard drives to ensure complete backup of the data.
RAID 30:
RAID 30 is also known as dedicated parity stripe array. it has RAID 0 and RAID 3's characteristics, the two RAID 3 disk (n = 3 disks) composed of arrays, using a dedicated parity bit, and then form a two disk RAID 0 array, extract data across the disk. RAID 30 provides fault tolerance and support a larger volume size. like as RAID 10, RAID 30 also provides high reliability, because even if there are two physical disk drives fail (one for each array), data is still available.
RAID 30 requires a minimum 6 drives, it is most suitable for non-interactive applications such as streaming video, graphics and image processing. these applications in order to handle large files, and require high availability and speed.
RAID 50:
RAID 50 array known as the distribution of parity stripe. with RAID 30 is similar, and it has RAID 5 and RAID 0 common characteristics. It consists of two sets of RAID 5 disk component (at least 3 each), each group has used the distribution type parity bit, and then formed into two groups of hard drives RAID 0, experimental data collected across the disk. RAID 50 provides reliable data storage and excellent overall performance, and support for larger volume sizes. even if the two physical disk failure (per an array a), data can be successfully recovered.
RAID 50 requires a minimum of 6 drives, it is most suited for applications requiring high reliability storage, high read speed, high-performance data transfer applications. These applications include transaction processing and there are many users accessing small files of the office applications.
RAID 53:
RAID 53 disk structure known as the efficient data transmission. the structure of the implementation of the data section with the Level 0 array, in which each segment is a RAID 3 array. its redundancy and fault tolerance with RAID 3. This requires a high data transfer rate RAID 3 configuration of the system useful, but it is expensive and inefficient.
RAID 1.5:
RAID 1.5 disk array is a new way, it has the characteristics of RAID 0 +1, and different is that its implementation requires only two hard drives. On the surface, the formation of the disk after the RAID 1.5, both have the same data . Of course, RAID 1.5 is a can not use disk space disk array model, therefore, two 80GB hard disk after the formation of RAID 1.5, and RAID 1 is the same, that is, the actual use of space is only 80GB, 80GB is that it also backup data. If the two separate hard drives were running them in the original system, is uninterrupted. but through the practical application, we found that if the two hard drives in separate runs, the data will cause a slight change in the reorganized again disk array, the data can not achieve full recovery, but subject to less disk data.
Since RAID 1.5 and RAID 1 have very similar effects, it is how the experiment striped RAID 0 read and write operate it? So far, we have not really material to prove the following hypothetical: that the disk array controller chip with advanced control functions, allows the two disks in both bands the same way of recording data, but need to read time, the controller has to tell the program to read bands from different hard disk and then read the different bands in order to achieve improved performance of the RAID 0 results.
In fact, for home users , RAID 1.5, or in most cases the realization of improved performance is, after all, with RAID 0 RAID 1.5 features. Therefore, the performance of RAID 1.5 we do for a comparison test. Here, let's take a look at RAID 1.5 Performance:
RAID Level RAID 0 RAID 1 RAID 0 +1 RAID 1.5
name of the band + band mirror image striping + mirroring
allow failure No Yes Yes No copy of the redundant copy of a copy of the type
hot standby operation can not be hard to be number one or more
2 4 2
available capacity to reduce the maximum and minimum capacity among middle
2 50% free disk a disk
high read performance (disk size) were written between the middle of the highest performance high
intermediate between the highest performance high
the middle of a row between writing a typical application of high
rapid trouble-free reading and writing small files to allow failure, random data is written to allow the failure of small files, continuous data transfer small files to allow failure, random data transmission
performance test:
hardware platform
CPU Intel Pentium4 3.0GHz (Northwood, 800MHz FSB)
Graphics Qiheng large silver home Ti4680 VIVO
Driver:
nVIDIA Detonator 4 V43.50
Memory KingMax DDR400 DDRRAM 256MBt2
unified memory parameters set in the DDR 400 AUTO mode
hard Barracuda V SATA 120G (7200rpm, 8MB Cache) t2
DiamondMax 9 PATA 80G (7200rpm, 2MB Cache) t2
Driver:
Intel IAA 3.0 With RAID
motherboard MSI (MSI) 875P NEO
DFI (Diamond) Lanparty PRO875
PS: onboard sound card, network card device drivers for all closed
:
Intel Chipset Software V 5.00
operating system, Windows XP Professional + SP1 English
DirectX DirectX 9.0
testing software software platform
ZDNet WinBench 99 2.0
Sisoft Sandra 2003
HDTach 2.61
this test, we used the ICH 5 of the Soft RAID Promise PDC20378 the RAID function and features formed the SATA-RAID 0 system, and the Lanparty PRO875 motherboard onboard Highpoint 372N established PATA RAID 1.5 were compared. formed PATA SATA disk arrays and disk drives used in an array of different types, so the test results are for reference only.
WinBench 99:
by WinBench 99 tests, we can understand the set up RAID after, whether it is commercial or high-end disk performance, disk performance, have been greatly improved. DiamondMax 9 consisting of two RAID 1.5 in business The most powerful disk performance, reached 21,200, while the high-end disk performance by the ICH 5 RAID 0 SATA disk array control of the lead-leading, performance is quite satisfactory.
To further compare the PATA RAID 1.5 SATA RAID 0 and the difference We made the following comparison test:
HighPoint PATA RAID 1.5
Promise 378 SATA RAID 0
two RAID modes by contrast, we found that using DiamondMax 9 Plus (DiamondMax 9 generation) hard disk in the RAID set the results were very impressive 1.5 system, the disk transfer rates up to 95KB / s, while the average transfer rate is close to 70KB / s. data transmission, the transmission rate volatility is relatively routine, transmission performance and stability. and SATA as good The new generation of disk interface, of course, also showed remarkable strength. through two Baracuda V (Barracuda 5th generation) formed in the disk RAID 0 system, the maximum transfer rate on the outstanding performance without RAID 1.5, but can also be more than 90KB / s, but the minimum transmission rate higher than a certain distance RAID 1.5, the average transfer rate Gengshi Bi RAID 1.5 better.
Sisoft Sandra 2003:
in Sisoft Sandra 2003 disk performance tests, we have adopted an empty disk partition as a test of the target disk capacity are 20G. formed by PATA RAID 1.5 system, its performance did not WinBench 99 test, so good, and the composition of the two control chip SATA RAID 0 far. < br> HighPoint PATA RAID 1.5
Promise SATA RAID 0
ICH 5 SATA RAID 0
HD Tach 2.61:
HighPoint PATA RAID 1.5
Promise SATA RAID 0
ICH5 SATA RAID 0 < br> In the HD Tach 2.61 tests, RAID 1.5 and then show glory, the average 47.8MB / s read speed for an average of only 31.1MB / s for SATA RAID 0 vanquish. But in terms of CPU utilization, SATA RAID 0 were significantly higher than PATA RAID 1.5 is much smaller, the smallest of which is controlled by ICH5 RAID 0 system, only 1.2%. But even the highest CPU utilization RAID 1.5 system is only 2.8%, for the mainstream today than 2GHz machine configuration, it is negligible.
in SYSMark 2002 tests, we found that after the whole set up RAID performance has not been improved so much imagination, but the performance in office will have a 5% increase. PATA hard drives and set up the RAID 1.5 SATA RAID 0 hard drive is not very different form of.
but in PCMark 2002 test, RAID 1.5 RAID 0 level than another, people have to admire.
Summary:
RAID is a kind of a long history of disk array technology. it is the birth not only to ensure that our important data in redundant disk arrays continue to ensure the safe long-term storage, or limitations, an important means to improve performance. So far, RAID technology has been applied to computer penetrated into all areas, while the home computer market, the developers had not considered RAID market, has also been a course of development for several years.
in the past, PATA hard disk of the RAID set in the home market is usually only RAID 0, RAID 1 and RAID 0 +1, but today we have not only witnessed the birth of a RAID 1.5, RAID 1.5 also feel the unique charm. and will soon become the mainstream SATA hard drives RAID 0 system compared to established, PATA hard drives set up The RAID 1.5 not significant weak point, on the contrary, RAID 1.5 PATA hard drive to improve performance is like and to ensure the most efficient way of data security.