The landlord disk array is quite troublesome to say here.

I. Introduction of RAID technical specifications

RAID technology mainly includes several specifications such as RAID 0~RAID 7, with different emphases. Common specifications are as follows:

RAID 0:RAID 0 continuously divides data in bits or bytes, and reads and writes data on multiple disks in parallel, so the data transfer rate is high, but there is no data redundancy, so it cannot be regarded as a real RAID structure. RAID 0 simply improves performance, and cannot guarantee the reliability of data. The failure of one disk will affect all data. Therefore, RAID 0 can't be used in situations with high data security requirements.

RAID 1: Data redundancy is realized through disk data mirroring, and backup data is generated on paired independent disks. When the original data is busy, the data can be read directly from the mirror copy, so RAID 1 can improve the reading performance. The unit cost of RAID 1 is the highest in the disk array, but it provides high data security and availability. When the disk fails, the system can automatically switch to the mirror disk for reading and writing without reorganizing the failed data.

RAID 0+ 1: Also known as RAID 10 standard, it is actually the product of the combination of RAID 0 and RAID 1 standard. While dividing data in bits or bytes and reading/writing multiple disks in parallel, each disk is mirrored to achieve redundancy. Its advantages are both the extraordinary speed of RAID 0 and the high data reliability of RAID 1, but the CPU utilization rate is also higher and the disk utilization rate is lower.

RAID 2: Distribute data in blocks on different hard disks, and the block unit is bits or bytes. An encoding technique called "Weighted Average Error Correction Code (Hamming Code)" is used to provide error checking and recovery. This coding technology requires multiple disks to store inspection and recovery information, which makes the implementation of RAID 2 technology more complicated, so it is rarely used in commercial environment.

RAID 3: It is very similar to RAID 2 because the data strips are distributed on different hard disks in the form of blocks. The difference is that RAID 3 uses simple parity and stores parity information on a single disk. If one disk fails, parity disks and other data disks can regenerate data; If the parity disk fails, it will not affect the use of data. RAID 3 can provide a good transmission rate for a large number of continuous data, but for random data, even-odd disks will become the bottleneck of writing operations.

RAID 4:RAID 4 also blocks data and distributes it on different disks, but the block unit is blocks or records. RAID 4 uses a disk as a parity disk, and every write operation requires access to the parity disk. At this time, parity disk will become the bottleneck of write operation, so RAID 4 is rarely used in commercial environment.

RAID 5:RAID 5 does not specify parity disks individually, but accesses data and parity information across all disks. On RAID 5, the read/write pointer can operate the array devices at the same time, thus providing higher data flow. RAID 5 is more suitable for small blocks and random data. Compared with RAID 5, the main difference between RAID 3 and RAID 5 is that every time RAID 3 transmits data, it needs to involve all array disks. For RAID 5, most data transfers only run on one disk and can run in parallel. There is "write loss" in RAID 5, that is, each write operation will produce four actual read/write operations, in which the old data and parity information are read twice and the new data and parity information are written twice.

RAID 6: Compared with RAID 5, RAID 6 adds a second independent parity block. Two independent parity check systems use different algorithms, so the reliability of data is very high, even if two disks fail at the same time, it will not affect the use of data. However, RAID 6 needs more disk space to allocate parity information, which has greater "writing loss" than RAID 5, so its "writing performance" is very poor. Poor performance and complicated implementation make RAID 6 seldom applied in practice.

RAID 7: This is a new RAID standard, which has its own intelligent real-time operating system and software tools for storage management. It can run completely independently of the host and does not occupy the CPU resources of the host. RAID 7 can be regarded as a storage computer, which is obviously different from other RAID standards.

In addition to the above standards (such as table 1), we can also combine various RAID specifications like RAID 0+ 1 to build the required RAID array. For example, RAID 5+3(RAID 53) is a widely used array form. By configuring a disk array, users can usually get a disk storage system that meets their needs more flexibly.

At the beginning, the RAID scheme was mainly aimed at SCSI hard disk system, and the system cost was relatively expensive. 1993, HighPoint Company introduced the first IDE-RAID control chip, which can build a RAID system by using relatively cheap IDE hard disks, greatly reducing the "threshold" of RAID. Since then, individual users have begun to pay attention to this technology, because the hard disk is the slowest and most unsafe device in modern personal computers, and the data stored by users often far exceeds the price of the computer itself. With relatively low cost, RAID technology can make individual users enjoy double disk speed and higher data security. At present, IDE-RAID control chips in PC market mainly come from HighPoint and Promise companies, and some from AMI companies (as shown in Table 2).

IDE-RAID chips for individual users generally only provide support for RAID specifications such as RAID 0, RAID 1, RAID 0+ 1(RAID 10). Although they are not technically comparable to commercial systems, the speed improvement and security they provide are enough for ordinary users. With the continuous improvement of hard disk interface transmission rate, IDE-RAID chips are constantly updated. The mainstream chips in the chip market have all supported ATA 100 standard, while the new HPT 372 chip of HighPoint Company and the latest PDC20276 chip of Promise can even support IDE hard disk of ATA 133 standard.

In today's increasingly fierce competition among motherboard manufacturers and increasingly high requirements of PC users, there are not a few manufacturers with RAID chips on motherboards. Users can set up their own disk array directly without buying RAID cards, and feel the speed of disk soaring.

2. The method to realize IDE RAID through hardware control chip.

In the RAID family, RAID 0 and RAID 1 are the most widely used in personal computers. After all, few individual users are willing to use four or more hard disks to build RAID 0+ 1 or other hard disk arrays, so here are only two RAID methods. We choose an upgraded KT7A-RAID motherboard that supports IDE-RAID function, and introduce the installation of IDE-RAID step by step. Upgrade KT7A-RAID integrated HighPoint 370 chip, supporting RAID 0, 1, 0+ 1.

Hard disk is indispensable for RAID. RAID 0 and RAID 1 have different requirements for disks. RAID 1 (mirrored) disk mirroring generally requires two (or more) hard disks to have the same capacity, while RAID 0 (striped) disks generally do not have this requirement. Of course, it is ideal to choose a hard disk with similar capacity or even identical performance. In order to facilitate the test, we chose two 60GB Seagate Cool Fish IV hard drives (Barracuda ATA,No. ST36002 1A). The system adopts Duron 750MHz CPU, 2× 128MB firewood wind gold bar SDRAM and GeForce2 Pro graphics card, which should be said to be a common configuration. We also want to know the system requirements needed to build RAID.

Creation of 1.RAID 0

first step

First, back up the data on the hard disk. Many users do not attach importance to backup, especially some careless individual users. Creating RAID is a dangerous operation for data. If you are not careful, you may destroy the data of the whole hard disk. This is especially true for RAID 0, which we introduced first. When RAID 0 is created, the data on all disks in the array will be erased, including the hard disk partition table. Therefore, it is necessary to prepare a Windows 98 startup disk with Fdisk and Format commands, which is also an important thing to pay attention to in this step.

Second step

Set the jumpers of the two hard disks as Master, and connect them to the IDE3 and IDE4 ports of KT7A-RAID respectively (controlled by the HighPoint370 chip on the motherboard). Since RAID 0 will rebuild the partition table of two hard disks, we don't need to consider the order of hard disk connection (we will see that this order is very important when creating RAID 1).

Third step

Set BIOS and turn on ATA RAID controller. We enter the integrated peripheral option in the BIOS of the upgraded KT7A-RAID motherboard and turn on the ATA 100 RAID IDE controller. The upgrade technology suggests changing the boot order to ATA 100 RAID. In fact, we found that this is not feasible in the process of system installation. Isn't there a partition hard disk to boot? So we still set the floppy drive as the first choice.

Fourth step

The next setup step is to create the core content of RAID 0, and we will introduce it to you in detail in a graphic way:

1. Restart the computer after setting the system BIOS, and the hard disk will not be reported again during the boot test.

2. Disk management will be taken over by HighPoint 370 chip.

3. The following are very critical HighPoint 370 BIOS settings. Press "Ctrl" and "H" at the same time in the HighPoint 370 disk scanning interface.

4. The first thing to do after entering the HighPoint 370 BIOS setup interface is to select "Create RAID" to create RAID.

5. Select RAID mode in "Array Mode". Here you can see the options of RAID 0, RAID 1, RAID 0+ 1 and Span. Here we select RAID 0.

6. when 6. After selecting RAID mode, it will automatically exit to the previous menu to select "Disk Drive". Generally speaking, it is good to enter directly.

7. The next setting is the stripe unit size, and the default value is 64kB. You can't ignore any special requirements.

8. Then there is the "Start Creating" option. Before you press the "Y" key, please carefully consider whether there is any important data on the hard disk. This is your last chance! Once you start creating RAID, all data on the hard disk will be erased.

9. After creation, specify a startup disk and select one.

Press "Esc" to exit, and of course press "Y" to confirm.

HighPoint 370 BIOS does not provide a function similar to "exit without saving", and it is irreversible after modifying the settings.

Step five

After restarting the computer again, we can see the words "Striping Array #0 # 0 (RAID 0)" on the screen. Insert the startup disk made before and start DOS. Open the Fdisk program, huh? Why is only one hard disk visible? Yes, RAID array has been regarded as a whole hard disk. For the operating system, RAID is completely transparent, so we don't have to worry about the management of RAID disks, which are all done by the control chip. Next, partition according to the method of ordinary single hard disk, and you will find that the capacity of this hard disk has "changed". Calculate carefully, yes, the total capacity is the sum of two hard disks! We can compare the reading and writing of RAID 0 to a zipper, which separates the data on the two hard disks and reads the data faster without wasting disk space. Don't forget to activate the primary partition after partitioning and formatting.

Step 6

Choosing an operating system is very confusing for us. HighPoint370 chip provides driver support for Windows 98/NT/2000/XP. Considering that the RAID function is aimed at relatively advanced users, we chose the English version of Windows XP Professional, which has better support for new hardware (the English version is mainly to facilitate the later Winbench test, and you can use the Chinese version of the operating system when using RAID). Windows 2000 is also a good choice, but the hardware support is obviously not as good as Windows XP Professional.

Step 7

For a computer with RAID, the installation of the operating system is different from that under normal circumstances. Let's take a look at the chart, which appears after Windows XP completes the first step of "file copy" restart. The installer will prompt "Press F6 to install SCSI devices or RAID disks" in English. This process is very short, and users tend to ignore the prompts at the bottom of the screen.

After pressing F6, the installation option appears. Select "s" to install the RAID control chip driver, and select "Enter" not to install.

Pressing the "S" key will prompt you to insert the RAID chip drive disk.

Type Enter, and the installer will automatically search the program on the drive disk, select the program with "WinXP", and then press Enter.

If the provided version is inconsistent with the built-in driver version of Windows XP Profesional, the installer will prompt the user to choose.

Press "s" to install the driver provided by floppy disk, and press "Enter" to install the driver included with Windows XP Professional. After pressing "S", you need to confirm again, and this time press "Enter" (this ... too many confirmations, hehe). The next step is normal system installation, which is no different from normal installation.

That's all the installation settings for RAID 0. Now let's talk about the installation of RAID 1. Compared with RAID 0, the installation process of RAID 1 is much simpler, and it is not destructive if it is operated correctly.

2. Creation of RAID1

Although it is completely different from RAID 0 in principle, the installation and setting process of RAID 1 is similar to RAID 0, and the main difference is the setting in HighPoint 370 BIOS. In order to avoid duplication, we only pay attention to this part of the settings:

After entering the HighPoint 370 BIOS, select "Create RAID" to create:

1. Click Enter in Array Mode, and select the second item "Mirror for data security (RAID 1)" in the RAID mode option.

2. Then, when selecting the source disk, we remind the user again: be careful not to make a mistake.

3. Then the choice of target disk, which is what we call mirror disk or backup disk.

4. Then start writing.

5. After the creation is completed, the BIOS will prompt to make a mirror image, which is a long process.

6. It took us about 45 minutes to make a 60GB image, so we created RAID 1.

RAID 1 will copy the data from the primary disk to the mirror disk, so you need to be very careful when building RAID 1. Don't confuse the master disk with the mirror disk, otherwise the result will be tragic. RAID 1 can be created on two hard disks with no data, or it can be added on a hard disk with an operating system installed, which is much more convenient than RAID 0 (except that the process of making a mirror image is long). After creation, we tried to unplug one of the hard disks, and the HighPoint370 BIOS issued a warning. Press "Esc", and the other hard disk takes the responsibility of the original disk, and all the data is intact.

For adding RAID 1 to the hard disk with the operating system installed, we suggest the following steps: open the control chip in the BIOS → start the operating system to install the HighPoint 370 driver → shut down the computer, connect the source disk and the mirror disk to port IDE3 and port 4 respectively → enter the HighPoint 370 BIOS to set RAID 1 (see the above steps) → restart the system to complete the creation.

We conducted a simple test on two RAID types. Although the test results of RAID 0 are somewhat puzzling, it still feels much faster than a single hard disk in actual use, especially when Windows XP Professional starts very fast and the progress bar flashes by. As for the instability of the transmission rate curve, we estimate that it has something to do with platform selection. After all, integrated chips are easily disturbed when doing such high data throughput work. But even so, we can still see that the data transmission rate of RAID 0 system has reached a very high level, once close to 60 MB/s. Compared with RAID 0, although the performance of RAID 1 system is not obviously improved compared with that of single disk system, we find that the working curve of RAID 1 is very stable and rarely fluctuates.

Let's take a look at the disk test results in Winbench99 2.0, which is clear at a glance.

For users and operating systems, RAID 0 and 1 are transparent and do not affect any operation, just like using a hard disk.

Thirdly, RAID is realized by software.

In addition to using RAID cards or chips attached to the motherboard to realize the disk array, we can also directly use software to realize the RAID function in some operating systems, such as Windows 2000/XP with built-in RAID function.

Before understanding the software RAID function of Windows 2000/XP, let's take a look at a function in Windows 2000-dynamic disk management.

Compared with the basic disk, dynamic disk is called volume set instead of the previous partition method, and its function is actually the same as partition, but there are the following differences:

1. The disk capacity can be changed at will.

Dynamic disks can change the disk capacity without restarting the computer, without losing data, while basic disks will lose all data if they want to change the partition capacity (of course, there are some special disk tool software that can change the partition without destroying data, such as PQMagic).

2. Limitation of disk space

Dynamic disks can be extended to discontinuous disk space in disks, or a volume set can be created across disks, and several disks can be combined into a large volume set. The partition of the basic disk must be a continuous space on the same disk, and the maximum capacity of the partition is of course the capacity of the disk.

3. Number of volume sets or partitions

There is no limit to the number of volume sets that a dynamic disk can create on the disk. A relatively basic disk can only be divided into four zones at most on a disk, and can only be divided into a main zone and an extended zone when using DOS or Windows 9X.

* It must be noted here that dynamic disks can only be used in Windows NT/2000/XP systems, and other operating systems cannot recognize dynamic disks.

Because most users' disks are basic disk types, in order to use the software RAID function, we must convert them into dynamic disks: control panel → administrative tools → computer management → disk management, and switch one of the windows to the disk list in the view menu. At this time, we can convert the selected disk into a dynamic disk through the right-click menu.

When you split a dynamic volume, you can see these types of dynamic volumes.

1. Simple volume: contains disk space on a single disk and has the same partition function.

(When there are two or more dynamic disks in the system and there is unallocated space on both disks, we can choose the following two ways to divide the volumes. )

2. Cross-volume: Cross-volume consolidates unallocated space from multiple disks into a logical volume.

3. Striped volume: combine the unallocated space on multiple (2 to 32) disks into one volume.

(If two dynamic disks in the system have the same capacity as mentioned above, we will see another partition method. )

4. Mirror Volume: Two identical copies of a volume, each on a hard disk. That is, we often say RAID 1.

When we have three or more dynamic disks, we can use a more complex RAID method -RAID 5. At this time, a new partition form will appear in the partition interface.

5.RAID 5 volume: equivalent to a striped volume with parity, that is, RAID 5 mode.

For most PC users, building RAID 0 is the most economical and practical array form. Here, only the construction of software RAID 0 is explained:

If you want to use software RAID 0 in Windows 2000/XP, you must first convert the disks to be included in the array into the dynamic disks mentioned above (it should be noted that in the default disk management interface of Windows 2000/XP, basic disks and dynamic disks cannot be converted, please refer to the above description). Here we try to use partition striping, which is the difference between software RAID and using RAID chips to build a disk array. We chose a 29GB partition to divide the stripe volume. When dividing stripe volumes, the system will need corresponding partitions, which means that other dynamic disks must have 29GB or more unallocated space. After the allocation of striped volumes is completed, two volumes with the same size will be merged by the system, and our formatting and other operations will be carried out on two disks at the same time.

After building RAID 0, we decided to test its hard disk transfer rate to determine the performance improvement of this software RAID. The platform on which we build software RAID is different from the hardware RAID platform in the last article. In order to ensure the performance of CPU and the realization of our software RAID, we adopted higher-end systems: Athlon XP 1700+, Samsung's 256MB DDR memory, and ASUS A7V266-E motherboard. Because software RAID has low requirements on hard disk specifications, we chose Seagate Cool Fish 60GB and Western Digital 1200BB 120GB.

In the second half of the transmission curve, we clearly see that the hard disk transmission rate of software RAID 0 has reached 60MB/s, which completely exceeds the transmission rate of any hard disk in the array, and the advantages of RAID 0 begin to appear. For users who pursue high performance, this should be their dream.

It should be noted here that in the Linux environment, we can also use the Raidtools tool to realize the software RAID function. This tool can make soft RAID 0, RAID 1, RAID 4, RAID 5 and other disk arrays. Before using Raidtools, we must first determine whether the Linux kernel currently in use supports MD. If the kernel you are using is 2.0.X and you haven't compiled it yourself, soft RAID is supported in most cases. If you are not sure, you need to compile the core yourself.

Although RAID function can bring us better speed experience and data security, it should be pointed out that most cheap IDE-RAID solutions on the market are essentially "semi-soft" RAID, which only integrates RAID control information into RAID chips, so their CPU utilization rate is relatively high and their performance is not very stable. This is also the reason why the performance of software RAID 0 in high-end systems can sometimes exceed that of "hardware" RAID 0 scheme.

For users, a high-performance IDE-RAID storage system requires either a relatively strong CPU computing power or an expensive RAID card. Therefore, disk array should still be regarded as a relatively high-end application. But for beginners, it is also a good choice to use a simple and cheap disk array to improve the availability of computer data or improve the storage speed. Of course, its performance is far from that of high-end systems.