Learn about RAID technology

In the past few years, from being a 'luxury' component only on mainframe systems, workstations, servers, RAID has been included in desktop computers in the form of simple integration. However, it is possible that buyers know their motherboards have RAID technology but not everyone knows how to use them effectively. This article introduces basic information about RAID as well as some usage experiences to power up your PC.

WHAT IS RAID?

RAID stands for Redundant Array of Independent Disks. Initially, RAID was used as a protection solution because it allowed writing data to multiple hard disks at the same time. Later, RAID has many variations that not only ensure data safety but also significantly increase the speed of accessing data from the hard disk. Here are five commonly used types of RAID:

1. RAID 0

This type of RAID is preferred by users because of the ability to improve the data exchange performance of the hard disk. Requires at least two hard disks, RAID 0 allows computers to write data to them in a special method called Striping. For example, if you have 8 data segments numbered from 1 to 8, the odd numbered sections (1,3,5,7) will be recorded on the first hard disk and even numbered segments (2,4,6) , 8) will be recorded on the second disc. For simplicity, you can imagine that you have 100MB of data and instead of putting 100MB into a single hard disk, RAID 0 will help put 50MB into each hard disk to reduce theoretical work time in half. From there you can easily deduce if there are 4, 8 or more hard disks, the speed will be higher. While it may sound attractive, in fact, RAID 0 still contains the risk of data loss. The main reason lies in the way of recording torn information so that the data does not lie completely on any hard disk and every time you need to access information (eg a certain file), the computer will have to synthesize the word hard disks. If a hard disk fails, the information (file) is considered unreadable and lost. Fortunately, with modern technology, hardware products are quite durable, so there are not many cases of such data loss.

It can be seen that RAID 0 is really suitable for users who need quick access to large amounts of data, for example gamers or people specializing in graphics and digital video.

2. RAID 1

This is the most basic form of RAID capable of securing data. Like RAID 0, RAID 1 requires at least two hard disks to work. Data is recorded on 2 identical drives (Mirroring). In case a drive fails, the remaining drive will continue to function normally. You can replace the damaged drive without worrying about lost information. For RAID 1, performance is not a leading factor, so it is not surprising that it is not the first choice for speed enthusiasts. However for network administrators or those who have to manage important information, RAID 1 systems are indispensable. The final capacity of a RAID 1 system is equal to the capacity of a single drive (two 80GB drives running RAID 1 will let the system see only one RAID 80GB drive).

3. RAID 0 + 1

Have you ever wished for an agile storage system like RAID 0, as safe as RAID 1? Definitely yes and obviously that desire is not just yours. Therefore, the RAID system of 0 + 1 was born, synthesizing the advantages of both 'seniors'. However, the cost for a system of this type is quite expensive, you will need at least 4 hard disks to run RAID 0 + 1. Data will be recorded simultaneously to 4 hard disks with 2 Striping accelerated drives and 2 backup Mirroring drives. These 4 drives must be identical and when put into RAID 0 + 1, the final capacity will be ½ of the total 4 drive capacity, for example if you run 4 80GB drives then the amount of 'visible' data is (4 * 80) / 2 = 160GB.

4. RAID 5

This is probably the most powerful form of RAID for office and home users with 3 or 5 separate hard disks. Data and backups are divided on all hard drives. This principle is quite confusing. We return to the example of 8 data segments (1-8) and now there are 3 hard drives. The data numbers 1 and 2 will be written to separate drives 1 and 2, their backups will be written to the hard drive 3. The numbers 3 and 4 are written to drives 1 and 3 with the corresponding backup write to drive 2. The numbers 5 and 6 write to drives 2 and 3, and the backup segment is written to drive 1 and then the sequence repeats, paragraph 7.8 is written to drive 1, 2 and the backup section written to drive 3 as the original. Thus, RAID 5 ensures both improved speed and high security. The final hard disk capacity is equal to the total amount of disk usage minus one drive. That is, if you use 3 80GB drives, the final capacity will be 160GB.

5. JBOD

JBOD is not really a mainstream RAID, but it has some features related to RAID and is supported by most RAID controllers. JBOD allows you to attach as many drives as you want to your RAID controller (of course, within the allowed port limits). They will then be 'aggregated' into a larger hard disk for the system to use. For example, if you plug in 10GB, 20GB, 30GB drives, through RAID controllers with JBOD support, the computer will recognize a 60GB drive. However, note that JBOD does not bring any other extra value: no performance improvement, no data security solution, just connectivity and aggregate capacity.

6. Some other types of RAID

In addition to the types mentioned above, you can also encounter many other types of RAID but they are not widely used but are limited to computer systems for their own purposes, such as: Level 2 (Error-Correcting Coding), Level 3 (Bit-Interleaved Parity), Level 4 (Dedicated Parity Drive), Level 6 (Independent Data Disks with Double Parity), Level 10 (Stripe of Mirrors, in contrast to RAID 0 + 1) , Level 7 (Storage Computer brand, which allows adding buffers for RAID 3 and 4), RAID S (the invention of the EMC group and used in their Symmetrix storage systems). Besides, there are some other variations, such as Intel Matrix Storage that allows RAID 0 + 1 to run with only 2 hard drives or RAID 1.5 of DFI on BMC systems 865, 875. They have many differences but Most of them are improved versions of traditional RAID methods.

WHAT DO YOU NEED TO RUN RAID?

To run RAID, you need at least one controller card and two identical hard drives. The hard disk can be in any standard, from ATA, Serial ATA or SCSI, it is best if they are the same because a simple rule is when operating in a synchronous mode like RAID, the overall performance of the system The system will be pulled down by the lowest drive if available. For example, if you took a 160GB drive running a RAID with a 40GB drive (regardless of 0 or 1), you might have wasted 120GB because the control system only considered them a couple of 40GB hard drives (except for the field). JBOD combination as mentioned). The key to the number of drives is the type of RAID you are going to run. The communication standard is not very important, especially between SATA and ATA. Some newer BMCs allow you to run RAID in a way that mixes both of these interfaces together. Typically, MSI K8N Neo2 Platinum or DFI Lanparty NForce4 series.

RAID controller (RAID Controller) is the place to gather data cables connecting hard disks in RAID system and it processes all data going through it. This controller comes in many different forms, from separate cards to integrated chips on BMC.

For PC systems, although not yet popular, it is advisable to choose to buy BMC with integrated RAID because this is generally one of the most obvious and cheap solutions to improve system performance, not to mention their data safety value. In case the BMC doesn't have a RAID, you can still buy a PCI controller card in the market at a very low price.

Another component of a RAID system is not required but is sometimes useful, which is hot swap disk drives. It allows you to replace hard disks that have problems while the system is operating without having to shut down the computer (simply unlock, remove the drive and plug in the new drive). This device is often used with SCSI hard drives and is quite important for server systems that require continuous operation.

The software is quite simple because most modern operating systems support RAID very well, especially Microsoft Windows. If you use Windows XP, adding RAID is quite easy. The most important is the driver, but it's great when they are included with the device. RAID installation can cause some problems if you are inexperienced but still have a solution in the later part of the article.

There are two situations that will occur when a user upgrades a system RAID. If the additional RAID system is only used for storage or high-speed information exchange, the installation is very simple. However, if you intend to use it as a place to install the operating system, the software will be very troublesome and have to reinstall all from 0.

RAID FOR YOUR OWN

1. Select RAID type

So you are determined to upgrade your system to new heights. But choosing the right RAID type is not as simple as you think. With conditions in Vietnam, you can choose a number of RAID solutions including 0, 1, 0 + 1 and 5. Where RAID 0, 1 is the most economical and usually available on most current BMC lines. RAID types 0 + 1 and 5 are usually only available on high-end, expensive models.

RAID 0 is definitely the best option, but it is also the most fragile. For example, if you use 4 hard disks in RAID 0, the data transfer rate can be more than 100MB / s. This is a very attractive number for any PC user. However, the possibility of losing data also increased by 4 times. A computer hard disk is a machine product that moves and gets 'aging' gradually after a long time of use (thankfully, the aging hard disk is quite long). In addition, grid malfunctions or control errors can also lead to disaster. Therefore, RAID 0 should not be used for long-term data storage, but it is the number one choice for high-speed temporary drives, for example, web database storage. And if you plan to use it for a long time, add some hard drives and switch to RAID 0 + 1. It is really ideal if there is abundant financial resources.

RAID 1 if run alone will have no effect but to create another drive just like the main drive. Ordinary users may not be interested in RAID 1, except for those who have to store and manage really important documents like servers that store customer or account information. If you use RAID 1, you should consider adding hot swap tables because it will help recover data more quickly (you can remove the drive and make a copy to the new drive while the system is in progress). work as usual).

RAID 5 is currently the # 1 option for all types of computers thanks to its ability to both fix and accelerate. If you plan to build a RAID system of 4 or more hard disks, RAID 5 is definitely the best solution.

Combined RAID types , such as RAID 0 + 1 or RAID 50 (5 + 0) often give characteristics of component RAID types, but you should consider and only use them if necessary because of the cost of the software. The event is quite high. We can summarize as in the table.

2. Select the hardware

The first is selecting the right components. Regarding the RAID controller chipset, you don't have many options because they are basically integrated on BMC. However, you need to pay attention to the following points.

Currently the integrated RAID controller usually consists of two main types: the controller chip attached to the BMC or available from the chipset. Common include:

Integrated chipset:
+ Intel ICH5R, ICH6, ICH7. These Southbridge bridges are included with the i865 / 875/915/925/945/955 series.
+ nVIDIA nForce2-RAID (AMD), nForce 3 Series (AMD A64), nForce 4 Series (AMD A64 / Intel 775).

External control chips: There are many different brands such as Promise Technology, Silicon Image, Adaptec, but most commonly there are two Silicon Image Sil3112 and 3114 lines.

The types integrated in the south bridge chip often have low latency, easy to use. However, the feature is often not much and the software is limited, sometimes 'abuse' of system resources for reading / writing tasks. Types using third-party chips, latency is often higher (negligible), software and features are somewhat richer, using minimal resources; Removable cards are easy to change and remove when needed. However, it is important to note that the Silicon Image Sil3112 has relatively poor compatibility, so data loss may be lost when switching to other RAID systems. Sil3114 and higher have overcome this error. NForce and ICH5,6,7 systems can exchange hard drives easily, their BIOS RAID is also smarter and is often able to identify pre-formatted RAID hard drive groups.

Some of the latest motherboards, such as DFI Lanparty NF4 SLI-DR, support RAID 5 as well. Regarding desktop communication, usually you only find PATA or SATA types and because you have to use multiple cables, SATA will be a smarter choice even without faster speeds and more technological improvements. If you're financially proficient, you might want to keep an eye on some relatively professional products that allow you to plug in more RAM to use as a huge buffer to increase speed significantly.

For hard drives, you should choose the type with the ability to transfer large data and fast access speed. Access Time (Access Time) specifies the time required for the hard disk to find the required data and the smaller the better. In addition, the hard drive should also have large buffers (8MB or more), some new models have 16MB of buffer capacity and technologies that allow significant performance gains (such as Seagate NCQ). The best hard drives should be the same because RAID will be affected if there is a slow drive or a small capacity according to the criteria of 'worm soup'.

3. Install RAID

Installing RAID in general is mainly based on the mainboard's BIOS, RAID Controller and almost nothing difficult.

After plugging the hard drive into the correct RAID location on the motherboard (refer to the product documentation), enter the BIOS of the BMC to turn on the RAID controller and specify the relevant ports (usually in the Integrated Peripherals section). .

After this operation, you will save the parameters and restart the computer. Pay close attention to the notification screen and press the correct key combination when required by the computer (can be Ctrl + F or F4 depending on the RAID controller) to enter the RAID BIOS.

For RAID BIOS, although each type has a different interface (refer to the accompanying documentation), you basically have to do the following:

+ Specify which hard disks will join RAID.
+ Select RAID type (0/1/0 + 1/5).
+ Assign Block Size: This is the key to greatly affect the performance of hard drive rigs running RAID. For Striping RAID, Block size also means Stripe Size. If this parameter is set to be incompatible with usage, it will cause memory waste and reduced performance. For example, if Block Size has a value of 64KB, then at least 64KB will be written to the drive in any case, even if it is a 2KB text file. So this value should approximate the average size of the files you use. If the hard drive contains many small files such as Word documents, you should leave the size block small, if you store multiple movies or music, large blocks will give higher performance (especially for RAID 0 systems).

Besides, Block size also has another function that determines where the file will be written. Go back to the example Block Size 64KB, if the file is smaller than 64KB, it will only be written to a hard drive in the RAID system and thus will not have any performance improvements. In another case, a file of 150KB size will be written to 3 drives with 64KB + 64KB + 22KB segments and the controller can read information from three drives at the same time allowing significant acceleration. If you select a block size of 128KB, that file will only be recorded in 2 128KB + 22KB drives. In fact, you should choose Block Size of 128KB for desktop computers unless you have specific needs.

After the controller has completely recognized the new hard disk system, you install the operating system as well as the RAID drive format. Windows XP is a wise choice.

Installing Windows in general is the same as normal but you need to prepare a floppy drive and driver-containing floppy disk for the RAID controller. Immediately after pressing the keyboard to enter the installation, you must pay attention to the text below the screen to install Windows to press F6 in time. Then wait a while and when asked, press S to put the RAID driver into installation.

The remaining steps work as you would with a normal hard disk installation.

After stabilizing the system, pay attention to installing additional utilities that control the RAID system to take advantage of extended features and sometimes even performance. You can list some programs like Intel Application Acceleration RAID Edition or nVIDIA RAID Manager .

General note:

If you have a drive full of data and want to set up RAID 0, you must format the drive and redo everything. So find yourself a suitable backup plan. If using RAID 1, this is not necessary.

Usually with a RAID 0 system, you should have a small hard drive to store the most important things in case something goes wrong.

When the computer restarts (especially when it is unusually booted), it may take a long time for the RAID controller to recognize the drives, maybe even strange sounds. out on hard disk muscle part. You do not have to worry because this is completely normal because the controller must synchronize the operation of all drives in the RAID group it manages.

RAID hard disk groups often consist of several hard drives operating side by side, so the heat output is quite large, not beneficial in the long term. Find a cooling solution if conditions permit to avoid unintended trouble.

RAID type

Number of hard disks

Security

End capacity

Performance

Data security

Price

0

1+

Normal

100%

Very good

Least

Very low

first

2

Good

50%

Rather

Good

Low

5

3+

Rather

(x-1) / x

Good

Good

medium

0 + 1

4,6,8

Good

50%

Good

Good

High

SUMMARY

The value that RAID brings to the system is undeniable - higher safety and performance depending on configuration. In fact, RAID 0 and 0 + 1 are the most popular in the home environment. RAID 0 is the fastest but also the most dangerous, just a problem is considered to be all over. While RAID 1, although providing the most information security, often gives users a sense of wastefulness (spending money on 2 hard drives with only 1 performance and capacity). RAID 5 offers high performance and safety, but the control device is usually quite expensive, not to mention the amount of hard drive spending is higher so few people care unless the job needs to be done. Therefore, some users turn to using SCSI drives to solve the performance / information security issue, but the cost for a good SCSI system may be even more expensive.

If you take a closer look, we will see that the IDE standard has many problems, for example, drives are not designed to run continuously (very important for server systems), current ATA cables are too busy. so when using multiple drives, it will lead to a tight case and in the worst case, the heat generated will lead to system malfunction. But with the growing technology and new standards like SATA, RAID will surely have a bright future and become the ideal companion for high-end personal computer systems.