What is a raid array - varieties and configuration. Creating a program RAID array on the Linux platform program to create RAID

By creating a virtual set of volumes or RAID and its further processing on the same efficient technology that is used for conventional discs or volumes. The main problem here is correctly creating a virtual RAID from the components of its discs, and often the situation occurs when there are disks (physical disks or images) included in the RAID, but the parameters of the disk array are partially or completely unknown.

Setting the correct parameters is critical when creating a virtual RAID and a successful recovery of data from it. There are certain methods for manual definition of RAID parameters (see the "Definition of RAID parameters"), but for their successful use, a fairly good level of knowledge of RAID technology, features of file systems and other technical aspects are needed. All this creates difficulties in determining the RAID parameters, even for professional R-Studio users.

To solve this task, the R-Studio developed a unique RAID parameter recognition technology, which allows you to determine the parameters of the disk array for any file system of the RAID volume system, even if they are absolutely unknown to the user. The use of this technology does not require any special knowledge of RAID volumes, however, understanding some key principles for recognizing RAID parameters in R-Studio will significantly increase your chances when recovering data on a disk array.

This article will also contain and further discussed on a specific example of a number of general provisions for using R-Studio automatically defining RAID parameters. In conclusion, we will tell about a number of additional actions necessary to determine RAID configurations in a number of complex data recovery cases.

General provisions
Each time an automatic recognition of RAID parameters should be followed by a number of general rules.

All RAID wheels or their images must be included in the RAID scheme. If there is no single disk, the RAID parameters will not be recognized (even if the RAID disk will be created instead. The missing disk object will be created. The only exception here is the backup disk without RAID data - such discs can be excluded from the RAID structure and this will not affect. To determining the parameters of the disk array.
Turning on the backup disks into a virtual RAID schema optional. As stated in the previous paragraph, when creating a virtual RAID, the backup disks are not required to determine its parameters. The inclusion of them in the RAID scheme will not affect the result, but will significantly increase the time required for RAID processing. If you know exactly that the disk is the backup and there is no RAID data on it, then it can be immediately excluded from the virtual RAID. Otherwise, leave all the disks, but in this case the duration of the process will increase.
Recognition of RAID parameters is a sufficiently long process, especially in the case of large RAIDs. Need to be patient. At the same time, R-Studio shows the progress of the operation, and you can approximately estimate the remaining time.
The process may not always be completed successfully. The result depends on many factors. The most important types of data stored on RAID and the degree of volume damage is the same. The degree of data compression is also very important: the smaller the data is compressed, the higher the chances of the fact that the RAID parameters will be correctly defined. For example, RAID volume parameters with Microsoft Office documents or OpenOffice documents, uncompressed graphic files (* .BMP), databases, etc. Have significantly higher chances of being recognized than the RAID parameters on which compressed graphic files are stored (* .jpg, * .tiff, * .png) and video files. Also on recognition of parameters can significantly affect the data remaining on the disks from previous discs.

If necessary, the RAID parameter recognition operation can be executed several times.

Basics of automatic recognition of RAID parameters
Consider the easiest case of automatic recognition of RAID parameters.

It is assumed that the user knows the basic principles of working with R-Studio. For more information, refer to the R-Studio documentation.

Fig.1. RAID components

2. R-STUDIO will start automatically recognizing RAID parameters, showing the operation of the operation. After completion, the RAID parameter detection dialog box opens (RAID Parameters Detection - Completed) and results will be displayed.

Fig.2. Recognized RAID schemes
Click on the image to increase it.

R-Studio will automatically select the best from recognized options for RAID parameters.

Please note that sometimes there may be a difference between the originals originally found and the final displacement option. This is a completely ordinary situation that does not affect the final result.

Fig.3. The difference between the originally found offsets and the final displacement option
Click on the image to increase it.

3. Click the Apply button in the RAID Parameter Detection dialog box - Completed (RAID PARAMETERS Detection - Completed), and the selected structure will be applied to a virtual RAID.

Fig.4. Restable RAID scheme
Click on the image to increase it.

4. Review the files to confirm that the RAID parameters were recognized correctly. If the file system is defined on the RAID, the R-Studio will show it. To make sure that the RAID parameters were recognized correctly, you can open the volume and view some files.

Fig.5. Viewed file on correctly recreated RAID
Click on the image to increase it.

Further steps: What to do if R-Studio cannot recognize the RAID parameters from the first attempt
There are many reasons for which R-Studio cannot correctly recognize the RAID parameters from the first time. In some cases, the disk array data may be damaged significantly, which makes it impossible to automatically recognize RAID parameters. However, if the data is still more or less saved, you should use certain methods that allow you to correctly determine the RAID parameters using R-Studio.

RAID scanning
Virtual RAID scan helps when the disk array parameters are recognized at first glance (more) correctly, but the file system on the recognized RAID is not determined. This is especially useful when the space on RAID disks before the displacement is filled with zeros. In this case, the displacement from the first attempt can be recognized incorrectly.

Fig.6. Incorrectly found offset RAID
Click on the image to increase it.

In this case, R-Studio may not determine the file system on the RAID. You will see a virtual RAID virtual RAID (DRIVES) panel, but there will be no any recognized file systems below.

Fig.7. File systems on recognized
Click on the image to increase it.

Scanning a virtual RAID will allow you to determine the file system. Select the Virtual RAID element (Virtual Block RAID) and click Scan (SCAN). After scanning, you will see the file systems found.

Fig.8. Found file systems after scanning Raidn
Click on the image to increase it.

Changing recognized RAID parameters
RAID parameters are recognized and selected so that the most reliable results are for the widest range of cases. There are also two additional parameters that, if necessary, can be changed: "Offset Search Performance" and "Raid Layout Search Performance)". When they change, the limits of the found displacements and RAID structures displayed in the results of detection are changed. The default shows only the most closest to reliable results. However, for more complex RAID schemes (for example, RAID 6), the default parameters can cut off correct displacements or RAID structures. You can expand the display of the found results Moving engine "Offset Search Performance" and "RAID LAYOUT SEARCH PERFORMANCE" search performance (RAID Layout Search Performance).

The default parameters were used in the figure below, and the offset was not recognized.

Fig.9. Displacements are not recognized when using default settings
Click on the image to increase it.

After displacing the engine "Offset Search Performance" search performance "In the Advanced Options) in the value" Quick (Fast) ", the correct displacement was found.

Fig.10. The correct displacement has been found after the "Offset Search Performance" parameter is reduced (Offset Search Performance) "
Click on the image to increase it.

In the same way, the "RAID LAYOUT SEARCH PERFORMANCE" (RAID LAYOUT SEARCH PERFORMANCE) "can be shifted in the same way to find a greater number of RAID structures.

But first of all, you must search with default parameters. For less complex RAID schemes, the Advanced Options (Advanced Options) parameters will result in too much results to be shown and the definition of the RAID circuit will take longer.

Manual choice of other offset and other RAID scheme
This can help when RAID data is damaged to such an extent that none of the above recommendations gives the desired result.

Start from offset with a maximum number of ratio (HITS), then select other offsets and repeat the detection process. After that, choose the found RAID structure with the greatest probability (probability). Continue the selection process until the correct RAID parameters are found.

Conclusion
Data recovery with a faulty RAID is a rather time-consuming task, and the recognition of the correct RAID parameters is the basis for its successful execution. Of course, the best in terms of data recovery is the case when the RAID configuration is known, i.e., for example, it was saved or recorded before the emergence of a malfunction. In other cases, the automatic RAID parameter recognition technology built into R-Studio allows to obtain quite correct results. This technology works with a high degree of reliability for simple disk arrays; Also in other non-trivial situations, it can be used as a fairly effective tool for finding offset and RAID structure.

Create 1,5-TB RAID array at home

The volume of information is raised by a rapid pace. So, according to the analytical organization IDC, in 2006, about 161 billion GB of information was generated on Earth, or 161 exam. If you submit this amount of information in the form of books, you will get 12 ordinary bookshelves, only the length of them will be equal to the distance from the ground to the Sun. Many users think about acquiring all the more capacious drives, the benefits of them are declining, and for $ 100, you can now purchase a modern Winchester for 320 GB.

Most modern motherboards have an integrated RAID controller with the ability to organize arrays of levels 0 and 1. So you can always buy a pair of SATA disks and combine them into the RAID array. In this material, the process of creating RAID arrays of levels 0 and 1 is considered, comparing their performance. Two modern SEAGATE BARRACUDA ES (Enterprise Storage) of the maximum capacity of 750 GB are taken as the tested.

A few words about the technology itself. A redundant array of independent (or inexpensive) disk drives (Redundant Array of Independent / Inexpass Disks - RAID) was developed in order to increase the fault tolerance and efficiency of computer storage systems. RAID technology was developed at the University of California in 1987. Its founded the principle of using multiple discs of a small volume, interacting with each other through special software and hardware, as a single container disk.

The initial design of the RAID arrays provided a simple connection of memory regions of several separate disks. However, later it turned out that such a scheme reduces the reliability of the matrix and practically does not affect the speed. For example, four disks combined into the matrix will get four times more often than one similar drive. To solve this problem, engineers from the Berkeley Institute offered six different levels of RAID. Each of them is characterized by a certain fault tolerance, the tank of the hard drive and the performance.

In July 1992, RAID ADVISORY BOARD (RAB) was created, which is engaged in standardization, classification and study of RAID. Currently, RAB has identified seven standard RAID levels. The excess array of independent disk drives is usually implemented using the RAID Controller board. In our case, hard drives were connected to an integrated RAID-controller of the ABIT AN8-ULTRA motherboard based on the NForce 4 Ultra chipset. To begin with, consider the capabilities offered by the chipset to build RAID arrays. NForce 4 Ultra allows you to create RAID arrays of levels 0, 1, 0 + 1, JBod.

RAID 0 (Stripe)

The bundle of discs, also known as RAID 0 mode, reduces the number of disks when reading and writing for many applications. Data is divided between multiple disks in the array so that read and write operations are carried out simultaneously for several disks. This level provides high speed of performing read / write operations (theoretically doubling), but low reliability. For a home user - probably the most interesting option that allows you to achieve a significant increase in the speed of reading and writing data from drives.

RAID 1 (Mirror)

Mirroring of discs known as RAID 1 is intended for those who want to easily reserve the most important data. Each recording operation is performed twice, in parallel. Mirror, or duplicated, a copy of data can be stored on the same disk or on the second backup disk in the array. RAID 1 provides a backup of the data if the current volume or disc is damaged or has become unavailable due to a hardware failure. Mirroring of discs can be used for systems with a high availability or for automatic data redundancy instead of a tedious manual information duplication procedure on more expensive and less reliable media.

RAID 0 systems can be duplicated using RAID 1. Beware and disk mirrors (RAID 0 + 1) provides higher performance and protection. The optimal method for reliability / speed ratio, however, requires a large number of drives.

Jbod.

JBod - This abbreviation is decrypted as "Just a Bunch of Disks", that is, just a group of disks. This technology allows you to combine into an array of various containers, however, the speed increase in this case does not happen, rather, even on the contrary.

At our integrated RAID controller NVIDIA RAID, there are other interesting features:

Defining a defective disk. Many users of multi-disc systems buy several identical hard drives to fully take advantage of the disk array. If the array fails, it is possible to determine the defective disc only by the serial number, which limits the ability of the user to correctly determine the damaged disk.

The NVIDIA warning disk system simplifies identification by displaying the motherboard on the screen with an indicating non-working port so that you know exactly what disk should be replaced.

Installing a backup disk. Disk mirroring technologies allow users to assign backup disks that can be configured as a hot reserve, protecting the disk array in case of failure. The general backup disk can protect several disk arrays, and a special backup disk can serve as a hot reserve for a specific disk array. Support for the backup disk, which provides additional protection over the mirroring, was traditionally limited to high-level multi-line systems. NVIDIA storage technology transfers this opportunity to PC. A special backup disc can be replaced defective until the repair is completed, which allows the support team to choose any convenient time to repair.

Morphing. In a traditional multidiscory environment, users who want to change the status of the disk or multidisk array must reserve the data, delete the array, overload the PC and then configure the new array. During this process, the user must pass a lot of steps to configure the new array. NVIDIA storage technology allows you to change the current status of the disk or array using one action called morph. Morphing allows users to update the disk or array to increase productivity, reliability and capacity. But more important is that you do not need to perform numerous actions.

RAID Cross Controller.Unlike competitive multi-tech (RAID) technologies, the NVIDIA solution supports both Serial ATA (SATA) and parallel ATA drives within one RAID array. Users it is not necessary to know the semantics of each hard disk, as the differences in their settings are obvious.

NVIDIA storage technology fully supports the use of a multi-disc array to boot the operating system when the computer is turned on. This means that all available hard drives can be included in the array to achieve maximum performance and protection of all data.

Restoring data "on the fly".If you fail, the disk mirroring allows you to continue working without stopping thanks to a duplicate copy of the data stored in the array. The NVIDIA storage technology goes to step further and allows the user to create a new mirror copy of the data during system operation, without interrupting user access and data applications. The recovery of data "On the fly" eliminates a simple system and increases the protection of critical information.

Hot connection.NVIDIA storage technology supports hot connections for SATA disks. In the event of a disk failure, the user can disable the defective disk without turning off the system and replace it with a new one.

NVIDIA user interface.Thanks to an intuitive interface, any user who does not even have experience with RAID can easily use and manage NVIDIA storage technology (also known as NVIDIA RAID). A simple "mouse" interface will allow you to quickly identify discs for configuring in an array, activate the bundle and create mirror volumes. The configuration can be easily changed at any time using the same interface.

Connect and configure

So, with the theory figured out, now consider the sequence of actions necessary to connect and configure hard drives to work in the RAID array 0 and 1 levels.

First connect discs to the motherboard. You need to connect discs to the first and second or third and fourth SATA connectors, since the first two refers to the primary (primary) controller, and the second pair to secondary (secondary).

Turn on the computer and go to the BIOS. Select an Integrated Peripherals item, then - RAID CONFIG item. Our eye appears the following picture:

We put the RAID Enable, then turn on the RAID for the controller where the discs connected. This picture is IDE Secondary Master and Slave, but we need to put Enabled in Sata Primary or Secondary, depending on where you have connected drives. Press F10 and leave the BIOS.

After rebooting, the RAID-Disk Configuration window appears to configure, press F10. NVIDIA RAID BIOS - exactly here and you need to choose exactly how to configure discs. The interface is very clear, just choose the desired discs, the size of the block, and that's it. After that, we will be prompted to format discs.

For the correct operation of the RAID array in Windows, you need to install the NVIDIA IDE Driver driver - it is usually available on a drive with drivers coming with a motherboard.

After installing the RAID drivers, it is necessary to initialize. Make it easy - right-click on the "My Computer" icon on the desktop, go to "Control - Disk Management". The service itself will be initialized and formatting discs. After passing through these procedures, the RAID array is ready to use. However, before installing, we recommend you to familiarize yourself with the full instruction that comes with the motherboard - everything is described in detail.

Seagate Barracuda ES Hard Drive was introduced in June last year. Winchester was designed to support storage solutions using the fastest growing applications - larger server servers, volume media materials, as well as various data protection schemes.

Barracuda ES has an SATA interface, a maximum container of 750 GB, and the spindle speed is 7200 rpm. Thanks to the support of Rotational Vibration Feed Forward (RVFF) technology, reliability has increased when working in closely standing multi-disc systems. It is also worth noting the Workload Management technology that protects the disc from overheating, which positively affects the reliability of the disks.

As already noted above, the drive is equipped with the SATA II interface, supports NCQ and has 8/16 MB cache. Also available 250, 400 and 500 GB options.

For testing, Seagate courtesyly provided two top drive st3750640ns with a capacity of 750 GB equipped with 16 MB cache memory. According to its technical specifications, Barracuda ES discs are an almost complete copy of ordinary desktop hard drives, and only more demanding to environmental conditions (temperature, vibration). Plus, there are differences in support of branded technologies.

Specifications:

Spindle rotation speed	7200 rpm
Volume of buffer
Average waiting time	4.16 ms (nominal)
Number of heads (physical)
Number of plates
Capacity
Interface	SATA 3 Gb / s, NCQ Support
Number of plates
Type of servo	built-in
Permissible overload when working (reading)
Permissible overloads when stored
Noise level	27 dBA ( idle mode)
Dimensions	147x101.6x26.1 mm
	720 grams

Appearance

So the drive itself looks like.

It is noteworthy that discs are distinguished both firmware and controllers - in one case, the St Micro chip is used, in the other - AGERE.

Includes a miniature jumper, which switches the mode of operation of the interface with 3 Gb / s by 1.5 Gb / s.

Testing

Test Stand Configuration:

CPU	AMD Athlon 64 3000+
Motherboard	ABIT AN8-ULTRA, NForce4 Ultra
Memory	2x512MB PC3200 Patriot (PSD1G4003K), 2.5-2-2-6-1T
Main hard drive	WD 1600JB, PATA, 8 MB Cash, 160 GB
Video card	PCI-Express X16 GeForce 6600GT Galaxy 128 MB
Housing	BIGTOWER CHIEFTEC BA-01BBB 420 W
Operating system	Windows XP Professional SP2

A few words about the cooling system. Winchesters are installed in the basket, which was cooled by one 92 mm fan Zalman ZM-F2. For comparison, the test results are comparable with three more hard drives: Ide Samsung SP1604N, 2 MB cache, 160 GB WD 1600JB, IDE, 8 MB cache, 160 GB, WD4000YR with a capacity of 400 GB, SATA, 16 MB cache, seagate 7200.10 with a capacity of 250 GB, SATA, 16 MB cache.

When testing, the following software was used:

FC Test 1.0 Build 11;
PC Mark 05;
AIDA 32 3.93 (included in it plugin for testing drives).

Due to the fact that the results of Seagate Barracuda ES are almost identical (the difference is stacked in the measurement error) The results of Seagate Barracuda 7200.10 750 GB, previously, it was decided not to include the results of testing a single drive, in order not to overload graphs with unnecessary information.

Test results in AIDA 32 3.93:

At the end of last week I bought components for the computer and ran into a number of problems when setting up the equipment. The new computer is designed to store databases in the office of one company, therefore I needed a RAID array. The budget was about 20,000 rubles, therefore collected on the AMD platform. The M4A88TD-M motherboard and two identical WD 500 GB hard disk. To configure the RAID, the HDD array connected to SATA0 and SATA1 connectors. Created an array RAID 1, combining hard drives with increased reliability and fault tolerance. When hard drives are mirrors each other. Recommendations described below are suitable for configuring RAID0, increasing the speed of discs.

The first, went to the BIOS. For my motherboard, pressing the DEL button when loading, for the boards of other firms can be F2. The SATA configuration settings switched IDE mode to RAID. I pressed the F10 to save the settings and rebooted the computer.

Second, you need to turn on theraid array. This is the first moment on which he fell into a stupor. In the instructions for the Asus motherboard, there is no word about it. During the download of the computer, I clicked Ctrl + f. Opened Option Rom Utility menu. Here I chose the second point by pressing 2.

In this menu, I clicked Ctrl + C to create RAID. Walking on items included the RAID MODE functions to the RAID1 position, opposite the drives of Y. Then twice pressed Ctrl + Y, entered the name of the RAID array and saved the set parameters. Released and rebooted the computer.

Now when the computer is loaded, it can be seen that the RAID1 array is connected in the system.

Third, determined the priority of the load queue from different devices. To do this, I had to enter the BIOS again. DVD drive, for him my RAID, and the last connected device, i.e. flash drives.

On the RAID array installed Windows 7. In principle, further tips are suitable for installing Windows XP, Vista, Server 2008 and Windows 8 to the RAID array. Before the installation started, I went from another computer to the ASUS website and downloaded AMD RAID DRIVER. The RAID driver uploaded to a flash drive, it is not necessary to insert the connector into the USB before selecting the hard disk partitions. The Windows image was on DVD. After that, it moved to the installation of the OS.

Fourth, used the driver from the flash drive when it reached the selection of the section. Inserted the USB flash drive, pressed the download and review.

In the pop-up menu, selected the driver directory, operating system and discharge. In my case Windows 7 64bit.

Windows installer detected the AMD AHCI COMPATIC RAID Controller driver. It was enough to see the hard disk section. Hug a flash drive from USB port.

Here I was waiting for the second snag, when Windows 7 is not installed. I chose the standard way to create, the entire volume of the disk installer defined as the main one. Click further and got a mistake. The installation program failed to create a new or find an existing system partition. Additional information and so on. When windosw is not installed due to the section, the solution is to make the disk breakdown on the partitions. Removed all sections. Pressed SHIFT + F10.

Fifth, pressing SHIFT + F10, called the command line. SHIFT + ALT Returns the English keyboard layout language in the Russian Distribution. Entered the diskpart, the call command to call the disk work utility. Next List Disk command. I saw two disks in the system: disc 0 - flash drive, disk 1 - RAID array. Selected disk 1 Select Disk command 1. Next entered the Create Partition Primary Size \u003d 131072, created a system section of 128 GB. The Create Partition Primary team is responsible for it. Size command to determine the size of the disk.

The second part of the disk defined the Create Partition Extended command section. I did not use Size to enable all the remaining space into the second disk. That in the future will create a logical disk.

Selected the first section Select Partition command 1. And the Active command is marked as active. After that, closed the command line window. Click the update button.

After updating the list of sections, I saw two disks of 128 GB and 337 GB. Selected the first section and pressed the Next button.

Hotly expected inscription Installing Windows ... Windows installation has passed as usual.

I did several times in three evenings. Some attempts were with errors, which increased the time. If you need questions, write in the comments. For example, you need to restart the computer after breaking the disk to new partitions if the flash drive was inserted before installing Windows. All the above described was repeated at a time to make sure of the algorithm of five points. Installing Windows 7 on RAID works, checked!

Setting the RAID array and install the latest versions of Windows operating systems

Many users after creating a software RAID array using the operating system tools begin to be interested in whether it is possible to install the OS Operations itself for this RAID and is it possible to transfer already pre-installed and filled with Windows created by RAID 1 array. There is a high probability that the hard drive itself with the operating system and all programs can be out of order, and not just the Winchester on which there is a set of files.

It should be noted that if the hard drive with the operating system and with all the files will stop its work in any company, it will suspend its activities immediately for a certain number of days. Therefore, the procedure for creating a RAID array with an installed operating system refers to the category of mandatory processes.

It is also worth taken into account that in addition to RAID 1 massif also has the ability to create RAID 0 array. The second allows you to significantly increase the speed of the operating system.

In this article, we will try to deal with how you can create a RAID 1 from two hard drives and install the operating system on it. At the same time, the motherboard may be any.

It should be noted that you can use the method of conventional launch BIOS and the application of RAID technology. You need to choose it to transfer it an already installed operating system using a different set of utilities. However, for a start, we will analyze how you can install Windows.

For the start, it is necessary to create the easiest way to the flash drive for bootable purposes.

In order to implement the RAID 1 assembly of the array, you need to be in two hard drives or SSD drives. It is very important to check whether the RAID technology is supported by a specific motherboard model. You can purchase a pair of hard disk with a capacity of 250 gigabytes.

They must be placed in the system unit correctly.

In order to configure the system with the RAID array, you should open the BIOS and select the corresponding section there. On each computer device model, they will be their own.

After turning on the computer device, you must log in via the BIOS and click on the Delete key.

In the window that appears, visit the Advanced section and open it. The list should select SATA Configuration.

You must set the RAID position in SATA MODE SELECTION.

HotPlug must be in the Enabled position for hard drives that we will merge in RAID.

In order for all settings to be preserved, press the F10 button. Then there will be a procedure for rebooting a computer device.

In order to go to the RAID configuration settings, the system offers our alert to press us on Ctrl + I. This action should be performed.

On the new page that appears, you must find the Create A Raid Volume point and click on it. After that, immediately click on ENTER.

On the new page you must enter the name of the RAID array. It can be any. To introduce the RAID name, you must click on the space.

After entering the name, click on ENTER.

In a new window, we need to choose RAID1 (Mirror). If this is not done, then all data will remain on another hard disk. The controller will turn out to be on another hard drive in automatic mode.

Settings are made using the use of arrows on the keyboard. After their holding, you must click on Create Volume and on ENTER.

A new window will be a warning about the consent of the user with the fact that its data may be lost. You need to click on yes using the Y button.

Thus, it turns out that the creation of RAID 1 is completed and you can exit by pressing the ESC button.

For the consent, you need to press again on Y.

After rebooting a computer device, you must go to the boot menu and select a flash drive from the priority download list. Here should be the RAID, which we created earlier.

Then you should select.

In a new window, after reading the license agreement, click on accept.

On the page that appears, you must select the second point.

Now it is noticeable that two hard drives that were combined into the RAID array, the system detects as a whole.

On the Connect page, you must select Skip this step.

In a new window, you need to use standard parameters.

You can create a local account and click on ready.

The importance of reliable storage is obvious to a user of any level. Especially now, when the stored data volumes are growing with threatening speed, regardless of whether these data are personal (collection of photographs and video recordings) or corporate (financial and project documentation, results of scientific research, etc.). One of the means to help to solve the problem of the reliability of data storage is based on the organization of the disk RAID array.

RAID Concept

Raid (English. Redundant Array of Independent Disks is an excess array of independent disks) (albeit more accurate, perhaps will be "free interpretation": an array of independent disks with an excessive resource) is a hardware or software subsystem in which stored data is distributed (often with duplication) for several hard disks (physical or virtual). The most effective, both from the point of view of reliability and in terms of performance, is the hardware RAID subsystem. However, the software implementation can also bring considerable benefits, and in Linux there are all the necessary components for organizing a software RAID array.

Different types of RAID arrays

Above it, it was noted that in addition to its main function - ensuring reliability of data storage - RAID can help increase productivity, sharing data into multiple threads for simultaneous recording into multiple disks. The implementation of the RAID subsystem in Linux is somewhat different from the generally accepted, but the logical division into several levels is saved.

At the level RAID 0. Two or more disks are used only to increase productivity, as data / write data is separated. Here, "redundancy" is actually absent.

Array RAID 1. It is the first level providing redundancy. This mode is often called "mirroring", since the data is duplicated on all the disks of the array. The degree of reliability increases, but the performance of the recording operation is reduced, since the entry of the same data is performed several times. For an organization RAID 1. Requires at least two disks.

Feature of the massif RAID 4. It is a separate disk to record information about the control of data. Thus, the venge sites of this subsystem are waiting periods when recording exactly on this disk. For this reason, it is recommended to use RAID 5. In all cases, except for those RAID 4. It is extremely necessary and justified.

In array RAID 5. During recording, data and information about controlling are separated. therefore RAID 5. It was considered the most efficient and economical level before the emergence of new developments in this area: Raid 5ee., RAID 6., RAID 10. and combined levels RAID 1 + 0, RAID 5 + 0, RAID 1 + 5. For the organization of Massiva RAID 5. Requires no less than three discs.

On the further development of the concept of RAID arrays can be found on the above pages Wikipedia. Of particular interest is the comparison of "RAID 10 Versus RAID 5 in Relational Databases" on the English-language page.

The RAID software support appeared in the Linux kernel, starting with version 2.0, although for practical use the first version was unlikely to be suitable: it was very limited, and contained a fair amount of errors. Starting with nuclei of version 2.4, the situation has improved, and modern RAID implementations in Linux are quite suitable for practical application.

Creating and configuring a RAID array

The first experiments with the creation of RAID arrays are recommended to be carried out in the virtual machine environment, for example, VirtualBox. This is more safe, besides, each user has a computer with two-three physical disks.

For a detailed consideration, the RAID-array of level 1 is selected, since it is the easiest from an architectural point of view and has the greatest redundancy (from the point of view of reliability) an array.

When creating a RAID array on several physical disks, you should pay special attention to the fact that the discs have the same size, and in the perfect version it is best to use the disks of one model.

So, first in VirtualBox, you must create a virtual machine for Linux (with version 2.6 kernel), select for it a suitable memory size and create three hard drive with the same volume (20 GB for each disk will be quite enough). After loading the Linux system (you can use any Live-DVD or its ISO image) to work the terminal emulator (text console).

To mark the partitions on the disk you can use the utility fDISK.but more convenient is her "heiress" cFDISK. With a pseudographical interface that you can run from the console by the following command:

CFDisk / Dev / SDA

After starting, you should create a section of the swap (for example, the size of 1GB), and the remaining space (19 GB) to give the root section. It is important to establish for both sections type Linux RAID (hexadecimal code fD.). After that, it is necessary to write down the changes made and exit cFDISK..

Now you need to perform exactly the same partitioning partitions on two other disks. This operation without difficulty is performed using another useful utility that allows you to manage disk sections:

SFDISK -D / DEV / SDA | SFDISK / DEV / SDB SFDISK -D / DEV / SDA | SFDISK / DEV / SDC

Key -d. Used to create a dump of all sections of the specified device in a format that can serve as input for the same command. sFDISK.that allows you to create duplicate partitions while saving all properties and attributes.

As a result, three disks with the same partitions and the installed Linux RAID type will be obtained. Now you can proceed directly to the creation of a RAID array.

Creating a RAID array disc

To create a RAID array will require a utility mDADM. From the same package. The creation operation itself is performed using the commands shown in Listing 1.

Listing 1. Creating a RAID array disc

MDADM --CREATE / DEV / MD0 --metadata \u003d 0.90 --level \u003d 1 --raid-devices \u003d 3 \\ / / DEV / SDA1 / DEV / SDB1 / DEV / SDC1 MDADM --CREATE / DEV / MD1 --Metadata \u003d 0.90 --Level \u003d 1 --raid-Devices \u003d 3 / Dev / SDA2 / DEV / SDB2 / DEV / SDC2

The first key of the command must define the main functioning mode mDADM.. In this case, the key is used. --Create. (brief form -C.) - creation mode. After this key, the name of the created RAID device is specified.

Key --metadata. (brief form -E.) Determines the format of metadata used for this disc (superblock). Value 0.90 (as well as its analogues 0 and default.) When creating a RAID array is used by default, so in the given example it would be possible and not included in the command. Format 0.90 Allows use in the RAID array to 28 components devices, and also determines the maximum size of these devices (no more than 2 TB).

Key --Level \u003d 1. Determines the level of the RAID array. Key --raid-devices \u003d 3 Determines the number of component devices in the RAID array created. Empitches the command line a list of devices from which the RAID array will be formed.

Running into operation

The disk array is successfully created, but it is still empty. You will need to install the Linux system to check the operation of the created RAID subsystem. Before installing, you should ensure that the section highlighted for a swap is correctly recognized by the installation program. To do this, use the following command:

mkswap / dev / md0 setup

Installing the Linux system is performed in the "regular" mode with the exception of one feature: when selecting sections instead of the usual / dev / sda1 and / dev / sda2 The user is offered / dev / md0 and / dev / md1 respectively. The loader, of course, must be installed in the MBR.

After the installation is completed, it is necessary to make changes to the loader configuration file so that the system is successfully loaded from the RAID device.

If the loader is used in the installed Linux system grub.then in the file /Boot/Grub/Grub.conf. The following changes must be made:

in line splashimage \u003d ... Section name (HD0,1) Replaced by name (MD0,1)
in line root (HD0,1) Similarly, replacement is performed on (MD0,1)

If the system is loaded using lilo. (This bootloader is still used in some distributions), it will be necessary to edit the file /etc/lilo.conf.:

add in the initial (general) section of the line: rAID-EXTRA-BOOT \u003d MBR-ONLY
replace string bOOT \u003d / DEV / SDA on a string bOOT \u003d / DEV / MD0

After saving and output from the editor, it is necessary to activate the changes made by the command lilo..

Now you can restart the system and check how the system on the RAID array works.

Accounting and control

The current state of the RAID subsystem in the kernel structure reflects the file / Proc / Mdstat. It should be noted that the authority of the superuser is required to view it. Information from this file is especially important in cases where a new disk is added to the RAID array (or a corrupted disk is replaced) or one of the component devices is removed from the RAID array.

Team mDADM. By itself performs the activation of the RAID array and, in principle, it does not require the configuration file, but will refer to it if it is exposed to its name ("Standard" name - /etc/mdadm.conf.). The use of the configuration file is recommended for two reasons:

disappears the need to write long command rows with a variety of keys, as all the characteristics are taken from the specified file;
the configuration file itself is the documentation for the RAID array used.

Team mDADM --Detail --scan. Allows you to get the values \u200b\u200bof the current RAID array settings. But to extract more details, you should use the next combination of commands (again, ROOT rights will be required):

Echo "Device / Dev / SD *"\u003e /etc/mdadm.conf mdadm --detail --scan \u003e\u003e /etc/mdadm.conf

RAID-array control

Advantages of redundancy When storing data in the RAID array, you can estimate if you disable one of the disks in the configuration of the virtual machine, thereby imitating it failure. After rebooting the system on the RAID array in a virtual machine, you must perform a copy operation of a sufficiently large amount of data. After that, the host system can be seen that the dimensions of the two running discs increased (in accordance with the amount of copied data), and the size of the third, disabled disk remained unchanged.

If the third disc is connected again and reload the system again in the virtual machine, the newly connected disk will be detected, but synchronization with the first two RAID disks will not be executed. The fact is that this operation must be performed manually. To continue the "experiment", you should now completely remove the third disk from the virtual machine and create exactly the same new one to simulate the replacement of a spoiled physical hard disk on a serviceable one.

The current status of the RAID array is verified by the contents of the aforementioned file. / Proc / Mdstat. After removing the third disk, the contents of this file will look approximately as in Figure 1.

From the shown fragment it is clear that only two RAID disks from three are currently functioning, and what kind of device component is missing, the parameter reports - The first two devices are involved, on the place of the latter - the symbol of the adhesion.

If we assume that the third disk in the virtual machine configuration is again created, then you must copy sections with all their characteristics from one of the working RAID disks using the above-described command:

SFDISK -D / DEV / SDA | SFDISK / DEV / SDC

After that, the prepared "new" disk is added to the existing RAID array:

MDADM / DEV / MD0 -A / DEV / SDC1 MDADM / DEV / MD1 -A / DEV / SDC2

Now the system continues to operate in normal mode - all three RAID components are connected and operate. You can make sure that you review the contents of the file again / Proc / Mdstat.

Additional comments

You can simulate the failure of one of the RAID discs using the Special Team Keys mDADM.As shown below:

MDADM / DEV / MD0 --SET-FAULTY / DEV / SDC1 or MDADM / DEV / MD0 --Fail / Dev / SDC1

After the disk is declared "non-working", it can be removed from the RAID array:

MDADM / DEV / MD0 --REMOVE / DEV / SDC1

Immediately after logically removing the device component, it can be replaced physically. Once again, it should be noted that "hardware" hard disks are highly recommended to replace only identical devices, and components in the form of logical partitions should certainly be replaced by sections of exactly the same size.

When the replacement of the component is carried out, you can add an updated component to the RAID array using the command mentioned above:

MDADM / DEV / MD0 --ADD / DEV / SDC1

Conclusion

The software implementation of the RAID array in the Linux system allows the user without any special difficulties to create disk arrays of several levels using both physical disks and logical partitions. Provided by this subsystem opportunity is enough to organize storage of data oriented on reliability and performance.