Raid is a RAID card in terms of hardware. Based on Different Hard Disk categories, raid cards can be divided into ide raid cards, SCSI raid cards, or SATA raid cards. Raid is named as a cheap disk in Chinese, mainly used for data recovery, performance improvement, and centralized hard disk management. RAID 0, RAID 1, and RAID 5 are the main types of raid. RAID 0 uses two SCSI hard disks. when data is written to the disk, the two hard disks start to write data at the same time. For example, 50% of data is written to both hard disks, the biggest advantage of this is to improve the performance, because two hard disks are simultaneously writing data of the same size. Theoretically, the performance can be improved by half, and the read performance is also the same, however, the disadvantage of this method is also obvious, that is, there is no data recovery function. Then, let's look at RAID 1. Assume there are two SCSI hard disks that are built using RAID 1. when data is written, data is written to both hard disks at the same time, and each Hard Disk needs to write all the data. In this way, the same data is written twice, and the data on the two hard disks is exactly the same, in this way, although the data recovery and backup functions are implemented, the performance is affected and disk space is wasted. Raid 2, 3, and 4 are not used much. I will not describe them here; RAID 5 is also a commonly used raid method. First, RAID 5 requires at least three hard disks, no matter how many hard disks are used, in the end, the capacity equivalent to a hard disk will be lost (this capacity is used for differential information for recovery ). ). First, we assume that there are three SCSI hard disks and RAID 5 is used for integration. Now, when data is written, the data written on Hard Disk 1 is "1" (part of all data), and the data written on Hard Disk 2 is "2" (the remaining part of all data ), in this case, the data written to Hard Disk 3 is "1 + 2". When Hard Disk 1 is damaged, you can get all the data of Hard Disk 1 ("1 + 2"-"2"). Similarly, when Hard Disk 2 is damaged, you can easily recover all data on this hard disk, when Hard Disk 3 is damaged, Hard Disk 1 and Hard Disk 2 can be easily restored. This is the safest data recovery solution. However, this will also lead to the above series of actions to be performed each time data is written, and the 1 + 2 action is generally complicated (because it is not simply to retain the sum of the two pieces of data, instead, we need to analyze the differences between the two pieces of data and save only the differences, so that we can avoid wasting space ). Now these are all completed by hardware, generally using Intel chips (Intel i960 ). However, the above only analyzes the situation of three hard disks. The following analyzes the situation of multiple hard disks. For example, there are five hard disks. According to the above theory, it should be hard disk 1, 2, 3, 4 respectively to store data 1, 2, 3, 4, hard disk 5 to store data 1 + 2 + 3 + 4, in fact not like this, because of the difference analysis of 1 + 2, the algorithm is already very complex, so if you want to analyze four or more hard disks at the same time, it is obviously impossible, the best way is to keep the difference between the two hard disks, so there is the following method: divide the five hard disks into five equal points. When there is data, record 1 in the first part of Hard Disk 1, record 2 in the first part of Hard Disk 2, record 1 + 2 in the fifth aspect of Hard Disk 5, and record 3 in Hard Disk 3, record 4 on hard disk 4, record 3 + 4 on the fifth part of Hard Disk 1, and so on. The basic idea is that I am not very familiar with it, the general principle is that a hard disk is both a specific Recorder of data and a recorder of differential analysis data. These five hard disks have specific data and different data, no matter which part is damaged, the other four parts can be restored, and the difference analysis only needs to be analyzed in two copies at a time. Data difference.
In addition to improving the performance (such as RAID 0) and data backup and recovery, the RAID card can also add a large number of caches to the card to further improve the I/O performance.
Next we will discuss the scsicard. For the scsicard, it is OK to master the following points:
1. The scsicard has its own control system and logic. It can perform read and write operations without occupying CPU resources (in our PC, all the read and write data must be delivered through the CPU, transmit data through the bus ). This is very important and can greatly improve the system performance. The first two SC statements of BTW and SCSI are small computers, which further illustrates this point.
2. The speed of SCSI is very high. Now it is around 15000 rpm, at least 10000 RPM, so the performance is very good.
3. If it is used together with the RAID card, if there is a SCSI interface on the raid card and there is also a SCSI interface on the system motherboard, try to use the RAID card interface, because if it is connected to the motherboard, if the motherboard is damaged, the hard disk cannot be used. Therefore, it is recommended to connect to the RAID card. Of course, if you build a PC on your own and buy a motherboard that supports SCSI interfaces, you do not need to perform raid. You can directly buy a SCSI hard disk to connect it to it and use it on your own. The performance will be good.
4. SCSI hard disks are available for hot swapping and non-hot swapping, and the price is similar. But hot swapping is actually for raid. In raid, hot swapping can be performed. If there is no raid, unplug the SCSI hard disk directly on the system. It is estimated that the system will die.
At last, the RAID card of SCSI is about 2000 or 3000, And the raid of IDE is only 400 or 500. For SCSI hard disks, The RAID card is usually about two thousand. If you buy a hard disk of about 9 GB, only 100 or 200 RMB, ^ _ ^
In addition, after raid is used, multiple hard disks can be managed in a centralized manner. That is to say, no matter how many hard disks constitute an array in the background, in the foreground (at the operating system level ), is a disk, You can partition it, and then use
