JP3412665B2 - Disk array device and control method therefor - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array device and its control method capable of minimizing a decrease in access performance that occurs during data restoration of a failed disk drive.

[0002]

2. Description of the Related Art A disk array device 2 has a plurality of disk drives 4A, 4B, 4C and 4D as shown in FIG. FIG. 1 shows the basic configuration of the disk array device 2. The disk drives 4A, 4B, 4C and 4D are controlled by the disk array controller 3 connected via the internal bus 5. The disk array controller 3 is also connected to the host computer 1 via the external bus 6.
Connected to each of the disk drives 4A, 4A in response to a read / write command from the host computer 1.
B, 4C, and 4D are accessed.

The disk drive 4A or the like normally stores information such as whether it is functioning normally or an abnormal state is detected as internal data. This storage information is checked by the host computer 1 and the disk drive 4A
Etc. are detected. Specifically, when the host computer 1 reads and writes data, the disk drive 4
When a read / write error (R / W error) is received from A or the like, the storage information of the disk drive 4A or the like is checked, read / write is retried according to the symptom, or the failure of the disk drive 4A or the like is determined. .

To detect a data error due to a failure of the disk drive 4A or the like, a method of providing a parity disk has been conventionally used. The method of providing this parity disk is, for example, processing each data of the same byte written in the same physical sector of each disk drive 4A etc. by exclusive OR operation and generating parity data based on this operation result. It is recorded together with the data.

However, with this type of method, it is impossible to restore data when the disk drive 4A or the like fails. Therefore, the disk array controller 3 of the conventional disk array device 2 includes a disk drive 4A
New disk drive 4 that has been replaced
A processing function of restoring the data of the failed disk drive 4A using the parity data is provided on A and the like.

This data restoration processing function is normally executed in order from the first sector on the new disk 4A,
And it is executed as a background process. The sector restored by such background processing does not affect the read / write command from the host computer 1 at all. That is, normal high-speed read / write processing is possible.

[0007]

By the way, according to the above-mentioned data restoration processing function of the disk array device, when a read / write command is received from the host computer while the data restoration processing is being executed, the restoration is performed. Data restoration processing is performed in the foreground processing for the unsectors. Therefore, there is a problem that the access performance as seen from the host computer is significantly reduced.

That is, in the case of a data write command, the data is written in the unrestored sector of the new disk drive, and the parity data is updated based on this data, so that almost normal access performance is maintained. In the case of a read command, which has a much higher execution probability than a write command, the desired sector is restored to the new disk drive and then the data is read from that sector. The problem of significantly lowering occurs.

The main objects to be solved by the present invention are as follows. It is a first object of the present invention to provide a disk array device and a control method therefor capable of minimizing a decrease in access performance that occurs during data restoration of a failed disk drive.

A second object of the present invention is that a file immediately after being accessed has a very high probability of being re-accessed.
It is an object of the present invention to provide a disk array device and a control method thereof that can minimize the influence of a disk drive failure in a multimedia service system including on-demand.

Other objects of the present invention include the specification, drawings,
Especially, it will be apparent from the description of each claim.

[0012]

[Means for Solving the Problems] To solve the above-mentioned problems,
The present invention achieves the above object by adopting the novel characteristic construction means and techniques listed below.

That is, the first feature of the device of the present invention is a disk array device having a plurality of disk drives for recording data and a disk array controller connected to the disk drives via an internal bus. When storing data in the disk drive, the array controller divides the data into a predetermined data length, and a redundant data generation function unit that generates redundant data for the divided data group, and the disk drive. A data restoration function unit that restores divided data that cannot be played back due to a failure by referring to redundant data, an access probability table that records the access probabilities of files recorded in multiple disk drives, and a restoration A repair table that sets a flag for files that should be, When writing data, the redundant data generation function unit generates divided data of a predetermined length and redundant data generated for the divided data, and the data group is stored in the plurality of disk drives. In addition to recording in a distributed manner, when the flag in the recovery table of the recorded file is reset, data is read and written, and the access probability is calculated every time a file recorded in a plurality of the disk drives is accessed. When the access probability of the corresponding file is recorded in the table and the abnormal state of the disk drive is detected and the failed disk drive is switched accordingly, the flag in the restoration table is set for the file to be restored. And set the file in the data restoration function section. The original process is performed, the restored data of the file is recorded in a new disk drive, the background process of resetting the flag in the restoration table of the restored file is performed, and the file read command is issued during the background process. When the file is received, if the flag in the restoration table of the file is set and the access probability of the file recorded in the access probability table is equal to or more than a predetermined threshold value, the file is restored in the data restoration function unit. Is performed, the restored data of the file is recorded in a new disk drive and the flag in the restoration table of the file is reset, while the flag in the restoration table of the file is set, and Record in access probability table When the accessed access probability is less than a predetermined threshold value, the data restoration function unit restores the file, and the read processing is performed without recording the restored data of the file in a new disk drive. To do
The disk array device is configured as described above.

A second feature of the device of the present invention resides in a disk array device in which the data length divided by the disk array controller in the first feature of the device of the present invention is a bit, byte or word unit.

A third feature of the device of the present invention is that the disk drive in the first or second feature of the device of the present invention is provided with a spare disk drive to be replaced when a failure occurs. On the device.

A fourth feature of the device of the present invention is that the disk drive in the first, second or third feature of the device of the present invention detects an abnormal state of the disk drive when the computer reads or writes data. The disk array device stores information for doing so as internal data.

A fifth feature of the device of the present invention is that the disk drive in the first, second, third or fourth feature of the device of the present invention provides a multimedia service such as video on demand. The disk array device is an external storage device of the host computer.

The first feature of the method of the present invention is that when data is written when controlling a plurality of disk drives by a disk array controller connected via an internal bus, divided data of a predetermined length and redundant data are written. Is generated, the data group is distributed and recorded in a plurality of disk drives, and if the flag of the recorded file is reset, the data is read and written, and the data is recorded in the plurality of disk drives. Each time a file is accessed, the access probability of that file is recorded,
If an abnormal condition of a disk drive is detected and the failed disk drive is switched accordingly,
A background that sets the flag for the file to be restored, restores the file, records the restored data of the file in a new disk drive, and resets the flag of the restored file. When a file read command is received during the background process, if the file is unrepaired and the access probability is equal to or higher than a predetermined threshold, the file is restored and the file is restored. The restored data is recorded in a new disk drive and the flag of the file is reset, while the file is unrestored and the access probability is less than a predetermined threshold, the restoration process of the file is performed. The recorded data of the file is not recorded on the new disk drive, and The control method for a disk array apparatus for carrying out the process.

A second feature of the method of the present invention is that the access probability in the first feature of the method of the present invention is obtained from the number of file accesses per unit time or from the elapsed time after the file is accessed. Another method of controlling a disk array device is described below.

The third feature of the method of the present invention is that the threshold value of the access probability of the file in the second feature of the method of the present invention is the number of accesses per average waiting time until the file is restored = 2. (Kw-1) / {TR (KE-1)}, where Kw = data read time with data recovery for writing / data read time when all disk drives are normal, KE = data for no writing Data read time accompanied by repair / data read time when all disk drives are normal, TR = time required for repair of new disk drive.

A fourth feature of the method of the present invention is that the data length divided by the disk array controller in the first, second or third feature of the method of the present invention is a bit, byte or word unit. It is in the control method of the array device.

A fifth feature of the method of the present invention is that the disk drive in the first, second, third or fourth feature of the method of the present invention is a spare disk to be replaced when a failure occurs. A method of controlling a disk array device provided with a drive.

A sixth feature of the method of the present invention is that the disk drive according to the first, second, third, fourth or fifth feature of the method of the present invention uses a computer to record an abnormal state of the disk drive. In the method of controlling a disk array device, information for detecting when reading and writing is stored as internal data.

A seventh characteristic of the method of the present invention is that the disk drive in the first, second, third, fourth, fifth or sixth characteristic of the method of the present invention is a video-on-demand type or the like. A method of controlling a disk array device, which is an external storage device of a host computer that provides multimedia services.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to its apparatus and method examples with reference to the accompanying drawings.

(Device Example) FIG. 1 shows a basic configuration of a disk array device 2 of this device example, FIG. 2 shows a flow of normal data read / write processing, and FIG. 3 shows data read at the time of failure. / Shows the flow of write processing.

The disk array device 2 has a plurality of disk drives 4A and 4A, as described in the section of the prior art.
B, 4C, 4D. Each disk drive 4A, 4
B, 4C and 4D are controlled by the disk array controller 3 connected via the internal bus 5. The disk array controller 3 is connected to the host computer 1 via an external bus 6, and can access the disk drives 4A, 4B, 4C, 4D in response to a read / write command from the host computer 1. Done.

In the disk array device 2 of this device example, when the host computer 1 issues a data write command to the disk array controller 3, the disk array controller 3 stores the data transferred from the host computer 1, for example, in FIG. As shown in (3), while dividing into three, for example, even parity data is generated, and a total of four divided data is generated. The data thus generated is stored in four disk drives 4
The data is divided into A, 4B, 4C, and 4D and recorded.

When the recorded data is read, the three divided data except the normal parity data are read from the three disk drives 4A, 4B and 4C, combined, and transferred to the host computer 1. .
For example, if one disk drive 4C fails here, the parity data and the divided data are read from the three disk drives 4A, 4B, and 4D that have not failed as shown in FIG. The divided data that cannot be played back due to the failure of the disk drive 4C is calculated, the restored data is written to a new disk drive (not shown), and the written data is read again to the host computer 1. Forward.

If the disk drive 4D recording the parity data fails, the parity data is restored in the same manner. However, since the data can be read out by the normal procedure, the host computer 1 can see it. Access performance does not deteriorate.

FIG. 4 shows the configuration of the disk array controller 3 in the disk array device 2 of this device example. The disk array controller 3 is the host computer 1
The redundant data generation function unit 31 that splits the data transferred from the disk drive and generates the parity data, detects the failed disk drives 4A to 4D, and performs the parity operation to perform the disk drives 4A, 4B, 4C, and 4D.
Data recovery function unit 32 for recovering the divided data that cannot be reproduced due to the failure of the above, an access probability table 33 for recording the access probability of the file recorded in the disk array device 2, and the file is recovered by the disk array controller 3. The recovery table 34 indicating whether or not the disk drive has been executed, the internal bus interface 36 for accessing the disk drives 4A, 4B, 4C, 4D, and the external bus interface 35 for connecting to the host computer 1.
When writing data, the redundant data generation function unit generates divided data of a predetermined length and redundant data generated for the divided data, and the data group is stored in the plurality of disk drives. In addition to recording in a distributed manner, if the flag in the recovery table of the recorded file is reset, the data is read and written, and the access probability table is written each time a file recorded in multiple disk drives is accessed. When the access probability of the corresponding file is recorded and an abnormal state of the disk drive is detected, and the failed disk drive is switched accordingly, a flag in the recovery table is set for the file to be restored, and The data restoration function section restores the file and The restored data of is recorded in a new disk drive, the background process of resetting the flag in the restoration table of the restored file is performed, and when a file read command is received during the background processing, the restoration table of the file If the flag is set and the access probability of the file recorded in the access probability table is greater than or equal to the predetermined threshold value, the data restoration function unit restores the file and restores the restored data of the file. On the new disk drive to reset the flag in the repair table for the file, while the flag in the repair table for the file is set, and the access probability recorded in the access probability table is the predetermined threshold value. Data recovery if less than And restore processing of the file in function unit, the restored data of the file without recording the new disk drive, and is configured to perform a read process.

The redundant data generation function unit 31 and the data restoration function unit 32 in the disk array controller 3 are in the normal state where the disk drives 4A, 4B, 4C, and 4D have not failed, and the process flow shown in FIG. The data is recorded and reproduced in accordance with. And the disk drives 4A, 4
If one of the B, 4C, and 4D fails, the disk drive connected in place of the failed disk drive 4C, or the disk drive 4C in which the hot standby disk drive that was previously activated has failed
Restore data equivalent to.

(First Method Example) Next, an example method of the present invention executed by using the above apparatus example will be described. FIG. 5 shows the disk array device 2 of the first method example.
6 is a flowchart of a procedure of data restoration processing in the control method of FIG.

First, the failure of the disk drive 4C is detected, and it is determined whether the failed disk drive 4C is manually replaced or switched to the hot standby disk drive (ST1). When the failed disk drive 4C is switched, the disk array controller 3 identifies the disk drive 4C to be restored (ST2) and restore table 34.
Flags of all files registered in are set (ST3). It should be noted that the file in which the flag is set indicates that the file has not been restored, and will be reset only when restored.

Next, the disk array controller 3
The access probability table 33 is referred to (ST4), and the file with the highest access probability is extracted from the files to be restored (ST5). Subsequently, the data restoration function unit 32 is used to execute the restoration processing of the divided data in which the file cannot be reproduced, and the restored divided data is recorded in a new disk drive (ST6).

Then, when the file restoration process is finally completed, the disk array controller 3 resets the flag of the corresponding file in the restoration table 34 (ST.
7). The file thus restored and the flag of the repair table 34 is reset is subjected to the normal read / write processing shown in FIG. 2 for subsequent access.

Further, when it is determined that there is a file to be restored (ST8), the processes of ST4 to ST7 are performed again. That is, the disk array controller 3 continues to extract the file with the highest access probability from the files in which the flag is set in the repair table 34, performs the restoration process, and restores the file when there are no files to be restored. The process ends. All of these processes are executed in the background process.

On the other hand, when it is determined that the failed disk drive 4C has not been replaced (ST1), the restored divided data is not saved, so the disk array controller 3 simply restores the restored divided data and the reproducible divided data. The data is combined and transferred to the host computer 1.
Various methods can be considered for calculating the access probability, for example, the number of times a file is accessed per unit time or the elapsed time after the file is accessed can be used, but it is preferable to select it according to the characteristics of the system.

Next, the processing procedure when there is a read command from the host computer 1 while the file restoration is being processed in the background will be described. FIG. 6 shows a read procedure during file restoration in the control method of the disk array device 2 of the first method example, and FIG. 7 shows a write procedure during file restoration in the control method.

When a read command is issued from the host computer 1 while the file restoration is being processed in the background, the disk array controller 3 refers to the restoration table 34 as shown in FIG. It is confirmed whether the data has been restored (ST10). After deciding whether the desired file has been restored in this way,
If the desired file is restored, the desired file is read as shown in FIG. 2 (ST13).

On the other hand, if the desired file is not restored, the desired file is restored in the foreground (ST11), and the restoration table 3 for the restored file is executed.
The flag of No. 4 is reset and the desired file is read (ST12, ST13). Therefore, the data restoration function unit 32 restores the unreadable data due to the failure of the disk drive 4C, writes the data in a new disk drive, and then executes the normal read processing.

When a write command is issued from the host computer 1 during the file restoration process, as shown in FIG. 7, data is normally written in a new disk drive (ST20) and the written file restoration table is written. The flag of 34 is reset (ST21).

In this way, the disk array controller 3 restores the files in descending order of access probability,
Control the restored files so that normal access is immediately performed. If the host computer 1 issues a file read / write command during file restoration, the file is restored and then the file is read, or the file is immediately written, and then the normal operation is immediately resumed. The flag of the repair table 34 is reset so that it can be accessed. As a result, even when the disk array device 2 is performing the data recovery process of the disk drive 4C, the probability that the access performance will be lowered becomes extremely low.

(Second Method Example) Next, a second method example of the present invention will be described. FIG. 8 is the second
9 shows a flowchart of a procedure of data restoration processing in the control method of the disk array device 2 of the method example.

In this example of the method, first, similarly to the example of the first method, it is judged whether or not the failed disk drive 4C has been manually replaced or switched to the hot standby disk drive (ST1 ').

When the failed disk drive 4C is switched, the disk array controller 3 identifies the disk drive 4C to be restored (ST2 ') and sets the flags of all the files registered in the restoration table 34. Yes (ST3 '). It should be noted that the file in which the flag is set indicates that the file has not been restored as in the case of the first method example, and the file is reset only when restored.

Next, the disk array controller 3 extracts the files to be restored, for example, randomly or referring to the access probability table 33 in the order of high access probability (ST4 '), and uses the data restoration function unit 32. Then, the restoration processing of the divided data which cannot be played back is executed (ST5 '), and the restored divided data is written in the new disk drive 4C (ST6').

Finally, when the file restoration process is completed,
The disk array controller 3 resets the flag of the corresponding file in the restoration table 34 (ST7 ').
The file thus restored and the flag of the repair table 34 is reset is subjected to the normal read / write processing shown in FIG. 2 for subsequent access.

When it is determined that there is a file to be restored (ST8 '), the processes of ST4' to ST7 'are performed again. That is, the disk array controller 3
Continues to extract files to be restored from the files whose flags are set in the repair table 34 as appropriate,
Similarly, the restoration process is performed, and when there are no more files to be restored, the file restoration process ends.

All of these processes are executed in the background process. It should be noted that the method of calculating the access probability may be selected according to the characteristics of the system as in the case of the first method example described above.

Then, during the file restoration process, a read / write command (R / W command) is issued from the host computer 1.
A processing procedure when there is such a case will be described. FIG. 9 is the second
9 is a flowchart of a read / write procedure during file restoration in the control method of the disk array device 2 of the method example.

While the file restoration is being processed in the background, the read / write command (R
/ W command) is determined (ST9 '), and R /
When there is a W command, the disk array controller 3 refers to the restoration table 34 as shown in FIG. 9 and confirms whether or not the desired file is restored (ST.
10 ', ST11').

If it has been restored, the normal read / write processing shown in FIG. 2 is executed (ST12 '), and if it has not been restored, it is further judged whether or not it is a read instruction (ST13'). In the case of a read command, the access probability table is further referred to (ST14 '), and it is determined whether the access probability of the file is equal to or more than a predetermined threshold value (ST15').

If it is equal to or greater than the predetermined threshold value, the file restoration process is executed in the foreground process (ST16 '), and the restored file data is written to a new disk drive (ST17'). Then, the flag of the restoration table 34 of the restored file is reset (ST18 ') and the normal file reading process is executed (ST19').

On the other hand, when the access probability of the file is less than the predetermined threshold value, the file restoring process is executed (ST20 ') and the file reading process is executed (ST21').

That is, the restored file is transferred to the host computer 1 without writing the data of the restored file to a new disk drive and without resetting the flag of the restoration table 34 of the restored file. In this case, since the restored data is not recorded, when the file is accessed again, it is necessary to restore the same file again.

During the file restoration process at ST13 ',
When it is determined that the write command is issued from the host computer 1, the data is written to the new disk drive as usual (ST22 ') and the flag of the written file recovery table 34 is reset (S).
T23 ').

In this way, the disk array controller 3
Determines whether to record the restoration data of the read file according to the access probability of the file, and records the restoration data for a file with a high access probability,
Prevents an increase in read time during re-access.

Further, regarding the file having a low access probability, the restored data is not recorded and the background processing is entrusted to relatively suppress the increase of the read time. Therefore, even when the disk array device 2 is performing a data recovery process, the rate of deterioration in access performance is extremely low.

Next, in the control method of the disk array device 2 of this method example, the access time when writing data in the foreground process and the access time when writing data in the background process will be described.

(When Data is Written by Foreground Processing) FIG. 10 shows an access time when data is written by foreground processing using the disk array device 2 of this device example. When the file is read during the restoration process of the new disk drive and the restored data of the file is written to the new disk drive, the file read time is as shown in FIG.

Here, the increase rate of the first data read time for writing the restored data is Kw. Further, since the files are randomly restored, the average waiting time until the files are repaired is 1/2 of the time TR required to repair a new disk drive, and if a desired file is accessed N times during this time. Assuming that the average read time of the file in which the restored data is written in the new disk drive is {Kw + (N-1)} / N times the read time in the normal state of the disk drive.

(When Data is Written by Background Processing) FIG. 11 shows the access time when data is written by the background processing using the disk array device 2 of this device example. When the file is read during the restoration process of the new disk drive and the restored data of the file is not written to the new disk drive by the foreground process, the file read time is shown in FIG. Like

Since the writing of the restored data is not performed in the foreground processing, if the rate of increase in the read time due to the data restoration processing performed every time is KE, the average read time is the read time in the normal state of the disk drive. KE times that of. Therefore, the threshold value of the access probability is expressed by the following expression below under the condition that the above two average read times are equal.

[0065]   Access probability threshold = value of N at which average read times are equal /                           Average waiting time for files to be repaired                         = 2 (Kw-1) / {TR (KE-1)}

Specifically, as shown in FIGS. 10 and 11, K
If w = 2.0 and KE = 1.3, the threshold value of the access probability is 2 / (0.3TR). Write the restored data to a file that is expected to be read at least 7 times before the new disk drive is repaired, ie at least 3.5 times within the average wait time before the file is repaired For files that are expected to be read less than 7 times before a new disk drive is repaired, ie less than 3.5 reads at most within the average wait time before a file is repaired By not writing the restoration data, it is possible to minimize the deterioration of the access performance during the restoration processing of the file.

Although representative apparatus examples and method examples of the present invention have been described above, the present invention is not necessarily limited to the means of the apparatus examples and the method of the method examples.
The present invention can be appropriately modified and implemented within a range in which the object of the present invention is achieved and the effects described later are obtained.

[0068]

As described above, according to the present invention, the data recovery of a failed disk drive is efficiently performed based on the access probability of a file, so that the deterioration of the access performance which tends to occur during the data recovery is minimized. The effect is that it can be suppressed to the limit.

In particular, in a multimedia service system such as video-on-demand where the probability that a file immediately after access is re-accessed is extremely high,
It is suitable for minimizing the effect of disk drive failure.

[Brief description of drawings]

FIG. 1 is a block diagram showing a basic configuration of a disk array device.

FIG. 2 is an explanatory diagram showing a flow of normal data read / write processing by the disk array device of the device example of the present invention.

FIG. 3 is an explanatory diagram showing a flow of data read / write processing when a disk drive fails in the same apparatus.

FIG. 4 is a block diagram showing a configuration of a disk array controller 3 in a disk array device 2 of this device example.

FIG. 5 is a flowchart showing a procedure of data restoration processing in the control method of the disk array device of the first method example.

FIG. 6 is a flowchart showing a read procedure during file restoration in the above control method.

FIG. 7 is a flowchart showing a write procedure during file restoration in the above control method.

FIG. 8 is a flowchart showing a procedure of data restoration processing in the control method of the disk array device of the second method example.

FIG. 9 is a flowchart showing a read / write procedure during file restoration in the control method of the above.

FIG. 10 is a graph showing the access time when data is written in the foreground process using the disk array device of this device example.

FIG. 11 is a graph showing an access time when data is written in the background process using the same device.

[Explanation of symbols]

1 ... Host computer 2 ... Disk array device 3 ... Disk array controller 4A, 4B, 4C, 4D ... Disk drive 5 ... Internal bus 6 ... External bus 31 ... Redundant data generation function unit 32 ... Data recovery function section 33 ... Access probability table 34 ... Restoration table 35 ... External bus interface 36 ... Internal bus interface

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shigetaro Iwazu 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Nihon Telegraph and Telephone Corporation (72) Innovator Nobuyoshi Izawa 3-19, Nishishinjuku, Shinjuku-ku, Tokyo No. 2 Nihon Telegraph and Telephone Corporation (72) Inventor Keisho Shiramizu 2-9-1, Furo-cho, Naka-ku, Yokohama-shi, Kanagawa NTT Intelligent Technology Corporation (72) Inventor Kono Takashi 2-9-1, Furomachi, Naka-ku, Yokohama-shi, Kanagawa Prefecture NTT Intelligent Technology Co., Ltd. (56) Reference JP-A-5-314674 (JP, A) JP-A-6-208488 (JP, A) JP-A-8-137627 (JP, A) JP-A-6-266508 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G06F 3 / 06-3 / 08 G06F 12/00- 12/12 G11B 20/10-20/16

Claims (12)

(57) [Claims] 1. A disk array device having a plurality of disk drives for recording data and a disk array controller connected to the disk drives via an internal bus, wherein the disk array controller stores data in the disk drives. When storing, the data is divided into a predetermined data length, and a redundant data generation function unit that generates redundant data for the divided data group and divided data that cannot be reproduced due to a failure of the disk drive. A data restoration function unit that refers to and restores redundant data, an access probability table that records the access probabilities of files recorded on multiple disk drives, and a repair that sets a flag for the files to be restored If you have a table and The redundant data generation function unit generates divided data of a predetermined length and redundant data generated for the divided data, records the data group in a distributed manner in the plurality of disk drives, and records it. When the flag of the file in the restoration table is reset, data is read and written, and the access probability of the corresponding file is recorded in the access probability table each time a file recorded in a plurality of the disk drives is accessed. Then, when an abnormal state of the disk drive is detected and the failed disk drive is switched accordingly, the flag in the restoration table is set for the file to be restored, and the data restoration function is set. Part of the file is restored and the file The original data is recorded in a new disk drive, the background process of resetting the flag in the restoration table of the restored file is performed, and if a file read command is received during the background process, the restoration of the file is performed. If the flag in the table is set and the access probability of the file recorded in the access probability table is equal to or greater than a predetermined threshold value, the data restoration function unit restores the file, The restored data of is recorded in a new disk drive to reset the flag in the restoration table of the file, while the flag in the restoration table of the file is set and the access recorded in the access probability table. Probability is predetermined When it is less than the threshold value, the data restoration function unit restores the file, and the read process is performed without recording the restored data of the file in a new disk drive. A disk array device characterized by the above. 2. The disk array device according to claim 1, wherein the data length divided by the disk array controller is in units of bits, bytes or words. 3. The disk array device according to claim 1, wherein the disk drive is equipped with a spare disk drive that is used in the case of a failure. 4. The disk drive stores, as internal data, information for a computer to detect an abnormal state of the disk drive when reading or writing data, according to claim 1, 2 or 3. The described disk array device. 5. The disk according to claim 1, 2, 3 or 4, wherein the disk drive is an external storage device of a host computer that provides a multimedia service such as video on demand. Array device. 6. A disk array controller connected via an internal bus for controlling a plurality of disk drives, when writing data, generates divided data of a predetermined length and redundant data, and writes the data. The group is distributed and recorded on multiple disk drives, and if the recorded file flag is reset, data is read and written, and each time a file recorded on multiple disk drives is accessed. When the access probability of the file is recorded and an abnormal state of the disk drive is detected, and the failed disk drive is switched accordingly,
A background that sets the flag for the file to be restored, restores the file, records the restored data of the file in a new disk drive, and resets the flag of the restored file. When a file read command is received during the background process, if the file is unrepaired and the access probability is equal to or higher than a predetermined threshold, the file is restored and the file is restored. If the file is unrepaired and the access probability is less than the specified threshold while the restored data of the above is recorded in the new disk drive and the flag of the file is reset, the restoration process of the file is performed. The recorded data of the file is recorded on the new disk drive, and Carrying out de-processing method for controlling a disk array apparatus, characterized in that. 7. The method of controlling a disk array device according to claim 6, wherein the access probability is obtained from the number of times the file is accessed per unit time or from the elapsed time after the file is accessed. 8. The threshold value of the access probability of the file is the number of accesses per average waiting time until the file is restored = 2 (Kw-1) / {TR (KE-1)}, where Kw = Data read time with data recovery with writing / Data read time when all disk drives are normal, KE = Data read time with data recovery without writing / Data read with all disk drives normal The method of controlling a disk array device according to claim 7, wherein TR is a time required for repairing a new disk drive. 9. The method for controlling a disk array device according to claim 6, wherein the data length divided by the disk array controller is in units of bits, bytes or words. 10. The disk array device according to claim 6, 7, 8 or 9, wherein the disk drive is provided with a spare disk drive to be replaced and used when a failure occurs. Control method. 11. The disk drive stores information for detecting an abnormal state of the disk drive when a computer reads and writes data as internal data. 11. The method for controlling a disk array device according to 9 or 10. 12. The disk drive is an external storage device of a host computer that provides a multimedia service such as video on demand, and the disk drive is an external storage device.
7. A method for controlling a disk array device according to.