JPH09212311A - Disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the speed of a disk array device from being influenced as a whole by the delay of operation end of a disk device due to error correction. SOLUTION: A disk array controller 35 divides data transferred from a main computer 10 into prescribed units in parallel, successively writes the divided data in the same sectors of respective disk devices A31, B32, C33, generates parities corresponding to their data, and writes the parities in a disk device D 34. When a respone from one of the disk devices 31 to 33 is delayed at the time of reading out data, data from the disk device delaying the response are generated based upon the data read out from the other two disk devices and the parity data read out from the disk device 34.

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 in which a plurality of disk devices are simultaneously operated in parallel so as to respond to a request from a main computer by apparently responding to one disk device. It is about.

[0002]

2. Description of the Related Art Conventionally, in a disk array device, data is sent to the main computer side after all the operations of a plurality of disk devices in the disk array device are completed. Here, in the case of a disk array device having an error correction function, when an error occurs in some of the disk devices, if error correction is possible, error correction is performed and then data is transferred to the main computer. In this case, the main computer can continue to operate without being aware of an error in the disk array device.

The advantage of such a disk array device is that
There are three main points. First is the realization of large capacity. Although the limit capacity of a disk device is constantly changing with technological progress, there is a limit capacity for one disk device at any given time. For example, the limit capacity at the time of filing of this application is about 4 G (giga) bytes,
Further, when it becomes a mass-produced product, its capacity is about 2 GB. Since the above-mentioned disk device having a limit capacity of 4 GB cannot be mass-produced, it is generally expensive and low in reliability.

On the other hand, since mass-produced products are stable, they are generally inexpensive and highly reliable. For example, in a disk array device including three disk devices, if a mass-produced 2 GB disk device is used for each disk device, an apparent 6 GB hard disk is completed. Since mass-produced disk devices can be collected to create a disk device having a capacity larger than the limit capacity, a highly reliable and large-capacity disk device can be obtained. Further, if a disk device of 4 Gbytes having a limit capacity is used, it becomes possible to make a disk device of 12 Gbytes which cannot be manufactured at that time.

The second reason is improvement in processing speed. Since the disk device is a device having many physical operation parts, it is not easy to improve the operation speed. This is because there are many physical factors such as the rotation speed of the disk, the moving speed of the read / write head, and the time required for the head position to stabilize after moving the head. However, in the disk array device, since data that should originally be written in one disk device is distributed and written in a plurality of disk devices, when considered in total, the frequency of reading and writing to one disk device and reading Since the amount of data is reduced, the processing speed can be improved as a whole.

[0006]

By the way, in the conventional disk array apparatus, when a request for reading or the like is received from the main computer, the operation of the plurality of disk apparatuses to be accessed is completed. Waiting, I sent the data from them to the main computer. Here, in the case of a disk array device having an error correction function, if an error occurs in some of the disk devices, if the error correction is possible, the error is corrected and then the data is transferred to the main computer. .

However, when an error occurs, the error will be known by the following procedure. In other words, each disk device tries to read data in response to a read command, but if it cannot read the data, it may automatically execute retry read processing, and eventually a retry over may result in an error. Many. For this reason, a disk device in which an error occurs often takes more time to complete its operation than a disk device that normally ends.

For this reason, in the case of the conventional disk array device, even if one of the disk devices in the device has an error, even if the error is corrected and the data can be restored as a result, the disk can be recovered. This will affect the operation speed of the entire array device. In particular, there is an inconvenience that the response to the command from the main computer is lowered.

The present invention has been made in order to solve the above-mentioned problems, and the number of disk devices whose operation is relatively delayed due to the occurrence of an error may be within a permissible range in terms of error correction. For example, it is an object of the present invention to prevent the delay of the operation end of the disk device from affecting the speed of the entire disk array device.

[0010]

SUMMARY OF THE INVENTION In order to achieve this object, the invention according to claim 1 controls a plurality of disk devices by a disk array controller and apparently makes an access from a main computer. In a disk array device that is configured to respond by pretending to be one disk device, and further has an error correction function for data error occurrence of the disk device, the disk array controller responds to a request from the main computer. When a read command is issued to a plurality of disk devices, if a response from a part of the plurality of disk devices does not reach a predetermined time after issuing the command, the disk that does not respond Data corresponding to the device is created by the error correction function, and stored in the main computer. It is a disk array device according to claim that is configured to deliver the requested data.

In the disk array device of the present invention, a plurality of disk devices can be controlled by the disk array controller, and an access from the main computer can be made to appear to be one disk device and can respond. As described above as an advantage of the disk array device in the description of the prior art, it is possible to realize a large capacity and an improvement in processing speed. Further, since it is possible to correct the data by correcting the occurrence of the data error in the disk device, it is possible to improve the reliability.

In addition to such basic functions and effects,
In the disk array device of the present invention, when the disk array controller issues a read command to a plurality of disk devices in response to a request from the main computer, a response from some of the plurality of disk devices is a command. If the specified time has not passed since the message was issued, the data corresponding to the disk device is created by the error correction function without waiting for the response from the disk device, and the requested data is sent to the main computer. Send out.

This is because if there is no response even after the lapse of a predetermined time, the disk drive is executing read retry processing because it could not be read due to an error during the first read, so the operation end is delayed. Therefore, the data corresponding to the corresponding disk device is created by a predetermined error correction based on the data obtained from the other disk device without waiting for the end of the operation of the disk device. By doing so, even if there is a disk device whose operation is delayed due to execution of error retry, etc., it is possible to ignore it and create data to be transferred to the main computer by the error correction function. Therefore, the operation can be continued without affecting the operation speed of the entire disk array device.

In other words, in the case of the conventional disk array device, when an error occurs even in one of the plurality of disk devices, it takes longer than the disk device that normally ends by the processing such as read retry, etc. until the operation ends. Was overloaded. Therefore, even if the error is corrected and the data is repaired as a result, the operation speed of the entire disk array device is affected, and particularly the response to the command from the main computer is lowered.

On the other hand, according to the disk array apparatus of the present invention, the response from the disk apparatus which lowers the response is processed without waiting, so that the response of the entire disk array apparatus is not lowered. . As described above, when the data corresponding to the disk device that does not respond is created by the error correction function, the requested data is sent to the main computer, and then the next read request comes from the main computer again, It is also conceivable that the operation of the disk device that is executing the above-mentioned read retry or the like has not been completed. In such a case, the read command itself cannot be issued to the disk device, so as shown in claim 2, the read command is issued only to the remaining (presumably normal) disk device. It is conceivable to generate data corresponding to the disk device for which the read command could not be issued by the error correction function and to send the requested data to the main computer.

As one of such error correction functions, there is a parity check method. In that case, as described in claim 3, a predetermined number of data storage disk devices that operate in parallel and each parallel data recorded in a predetermined same storage position of each of the data storage disk devices. The disk controller is provided with a read command for a set of disk storages for storing parity stored in the disk drive, and the disk controller can execute error correction by the parity check method in units of the set of disk devices. It is possible to configure.

Here, the phrase "provides a read command for a set of disk devices consisting of a predetermined number of disk devices for data storage and disk devices for parity storage" means that the disk array device is, for example, three data storage devices. It is also possible to have a configuration in which only a total of four disk devices for storage and one disk device for parity storage are provided, but if these four devices are made into a set and a plurality of sets are provided, realization of large capacity and This is because it is more effective in improving the processing speed.

In this case, for example, a total of ten data storage disk devices and one parity storage disk device may be used, but there are the following problems. It is one of a total of 10 disk devices
If the response is slow for the machine, data can be created by the error correction function using the parity check method.
When the response is delayed for two or more units, the data for those two units cannot be created by the error correction function. As a result, the entire configuration of 10 disk devices cannot be used, which poses a large risk. Therefore, for example, a total of four disk storage devices including three data storage disk devices and one parity storage disk device are used.
By constructing a set of 3 units, the probability that the response of 2 units out of 4 units will be much slower than that of 2 units out of 10 units, which is also effective in improving reliability. .

In claim 3, the parity check method is used as an example of the error correction function, but the error correction function may be provided by other methods. .

[0020]

Next, an embodiment of the present invention will be described. A disk array device 30 according to an embodiment of the present invention shown in FIG. 1 includes four disk devices 31, 32,
33 and 34, and a disk array controller 35 capable of individually controlling the respective disk devices 31 to 34. In the following description, in order to distinguish four disk devices, the A disk device 31, the B disk device 32, the C disk device 33, and the D disk device 34.
Will be described. In addition, the disk devices 31 to 34
Is a so-called physical hard disk drive integrated with a control board for controlling it.

The disk array device 30 includes a main computer 10 and a SCSI (Small Computer System Interfac).
e) Connected via a standard bus 20, and the main computer 10 looks like a single disk device. The main computer 10 has a keyboard 11 as an input means and a CRT as a display means.
12 are connected.

In this disk array device 30, the disk array controller 35 has four disk devices 3
1 to 34 can be simultaneously operated in parallel. And
The disk array controller 35 divides the data transferred from the main computer 10 in parallel in a predetermined unit. In this case, the A disk device 31 and the B disk device 3 are divided.
2, the C disk device 33 is sequentially written on the same sector, and the parity corresponding to the data is generated and written to the D disk device 34.

Therefore, even if the data from any one of the A disk device 31, the B disk device 32, and the C disk device 33 cannot be read, the data from the other two devices and the D disk device 34. By exerting a predetermined error correction function based on the parity written in, the data from the disk device that could not read the data can be created (restored).

The present disk array device 3 having such a configuration
The basic advantages of 0 are mainly the following three points. First is the realization of large capacity. Although the limit capacities of the disk devices 31 to 34 are constantly changing with technological progress, the limit capacities of the disk devices 31 to 34 are always present. For example, the limit capacity at the time of filing this application is 4G
It is about (giga) bytes, and when it becomes a mass-produced product, its capacity is about 2 GB. Since the above-mentioned disk devices 31 to 34 having a limit capacity of 4 GB cannot be mass-produced,
Generally expensive and unreliable.

On the other hand, mass-produced products are stable and generally inexpensive and highly reliable. For example, as in the configuration of this example, an A disk device 31 for storing data,
In a disk array device 30 including three B disk devices 32 and C disk devices 33, if mass-produced 2 Gbyte products are used for the respective disk devices 31 to 33,
Apparently a 6GB disk device is completed. Since mass-produced disk devices can be collected to create a disk device having a capacity larger than the limit capacity, a highly reliable and large-capacity disk device can be obtained. In addition, when the disk devices 31 to 33 each having a limit capacity of 4 GB are used,
At that point, it is possible to make a 12 Gbyte disk device that is considered unmanufacturable.

Secondly, improvement in processing speed can be mentioned. Since the disk devices 31 to 34 are devices having many physical operation parts, it is not easy to improve the operation speed. This is because there are many physical factors such as the rotation speed of the disk, the moving speed of the read / write head, and the time required for the head position to stabilize after moving the head. However, the disk array device 30
Since a plurality of data, which should originally be written in one disk device, are distributed and written in the disk devices 31 to 33 of A to C in this example, in a total consideration, reading and writing to one disk device is performed. Since the frequency and the amount of data to be read are reduced, the processing speed can be improved as a whole.

Since the disk array device 30 of this example has three disk devices 31 to 33 for data storage and D disk device 34 for parity storage, which one of them is used? Data is not lost even if one unit is totally broken. The mechanism will be described with reference to FIG.

First, when writing data, the disk array controller 35 uses the A disk device 31 for data storage,
Data is sequentially written on three identical sectors of the B disk device 32 and the C disk device 33, the write data is subjected to exclusive OR operation processing, and corresponding parity data is generated based on the operation result. Then, the created parity data is written in the same sector of the D disk device 34.

As a concrete example, when writing data "101" as shown in FIG. 2A, "1" is stored in the A disk device 31 for storing data, and "0" is stored in the B disk device 32. , 1 is written in the C disk device 33, respectively. Then, when the write data is subjected to the exclusive OR operation, the parity data “0” is obtained as the even parity in the case of the present embodiment, so that it is D
Write to the disk device 34.

On the other hand, at the time of reading data, the disk devices 31 to 31
When a read error occurs due to a failure of 34, the disk array controller 35 controls as follows. It should be noted that in this example, it is not possible to cope with the case where an error occurs in two or more of the four disk devices 31 to 34, so the following description will be given assuming the case where an error occurs in any one of them. Proceed. When an error occurs in one of the disk devices 31 to 33 for data storage A to C,
Data read from the remaining normal disk unit and D
Based on the parity data read from the disk device 34, data corresponding to one disk device in which an error has occurred is created (restored) by the error correction function.

A specific example will be described with reference to FIG.
In the case of FIG. 2B, since the A disk device 31 has an error, the B disk device 32 gives “0”, the C disk device 33 gives “1”, and the D disk device 34.
From this, parity “0” is obtained as read data. When the above-described exclusive OR operation processing is performed based on these data, the data that should have been read from the A disk device 31 is "1" because the parity is even parity in this embodiment. Then, the created data is added and "101" is sent to the main computer 10 as normal read data.

Similarly, in the case of FIG. 2B, B
Since the disk device 32 has an error, the A disk device 31 obtains "1", the C disk device 33 produces "1", and the D disk device 34 obtains parity "0" as read data. When the data that should have been read from the B disk device 32 by the error correction function is created based on these data, it becomes "0", so that "101" including the created data becomes normal read data.

In the case of FIG. 2B, since the C disk device 33 has an error, the A disk device 3
1 to “1”, B disk device 32 to “0”, and D disk device 34 to obtain parity “0”, respectively, as read data. When the data that should have been read from the C disk device 33 is created based on these data by the error correction function, it becomes "1", so that "101" including the created data becomes normal read data.

In the case of FIG. 2B, the D disk device 34 has an error, but this is for parity storage in case an error occurs in the other disk devices 31 to 33. Therefore, there is practically no problem. That is, A disk device 31 to "1", B disk device 3
Since “2” to “0” and “1” from the C disk device 33 are transferred as read data, normal read data “101” can be obtained even if there is no parity data from the D disk device 34.

The configuration described above, and the operation and effect based on the configuration are known from the prior art.
Even assuming that, there are the following problems. That is,
When a read request is issued from the main computer 10 to the disk array device 30, the disk array controller 35 issues a read command to the disk devices 31 to 33 of A to C to be accessed, and all of them are read. Wait for all the operations of the disk devices 31 to 33 of the main computer 1 to finish, and then obtain the data obtained from them by the main computer 1
It was transferred to the 0 side. When an error that data cannot be read occurs in one of them, the parity data is read from the D disk device 34, and based on the data from the other two devices and the parity data of the D disk device 34. Creates data from a failed disk device using the error correction function. This created data is sent to the requesting main computer 10.

However, when a normal error occurs, after retry reading is automatically performed several times in the disk devices 31 to 34, eventually a retry over occurs and an error often occurs. For example, in the case shown in FIG. 2B, the A disk device 31 in which the error has occurred takes more time than the B and C disk devices 32 and 33 that normally end, until the operation ends.

Therefore, the B and C disk devices 32,
Even if the operation is completed in 33, the operation is completed in the A disk device 31, and it is not until the error is known for the first time that the process of error correction using the parity data of the D disk device 34 cannot be started. Therefore, if an error occurs in even one of the disk devices 31 to 33 for data storage in the disk array device 30, even if the error is corrected and the data is restored, the disk array This will affect the operation speed of the entire device 30. Especially, the response to the request from the main computer 10 is deteriorated.

On the other hand, the present disk array device 30 is also devised to prevent such a decrease in response. It is the main computer 10
This is the control when a data read request is issued from. The control will be described with reference to FIGS.
FIG. 3 is a flowchart of a data read control process executed when a data read request is issued from the main computer 10, and FIG. 4 is a disk array controller 35 and each disk device 31 to 31 by the control process.
It is explanatory drawing which shows the outline of operation | movement of 34.

First, in the first step S10 of FIG. 3, a read command is output to each of the disk devices 31 to 34 of A to D in response to a data read request from the main computer 10. After outputting the read command in S10,
If data is transferred from each of the disk devices 31 to 34 at 20, the data is stored in a buffer memory (not shown) in the disk array controller 35.

Then, in the following S25, it is judged whether or not the responses from all the disk devices 31 to 35 have been completed, and if the responses have been completed (S25: YES), the process directly proceeds to S50. The process of S50 will be described later. On the other hand, if not completed (S25: NO), S3
Move to 0.

In S30, it is determined whether or not a predetermined time has elapsed (whether a timeout has occurred) since the read command was output to each of the disk devices 31 to 34 in the process of S10. If not timed out (S30: NO), the process of S20 is repeated. If timed out (S3:
0: YES), and the process proceeds to S40.

In S40, it is determined whether or not there is no response among the disk devices 31 to 34 of A to D that output the read command in S10. The predetermined time set in S30 is a time sufficient to finish the data transfer after outputting the read command when the reading is normally executed. However, if the read retry is performed because the reading was not successful, this predetermined time may be exceeded.

Therefore, if the disk devices 31 to 34 are normally operating, a timeout occurs (S
Since there is no disk device that does not respond in 30: YES), a negative determination is made in S40, and the process proceeds to S50.
Of course, in the case of affirmative determination in S25 described above, since the response has already been completed in all the disk devices without waiting for timeout, the process proceeds to S50.

At S50, the read data is sent to the main computer 10. Affirmative judgment in S250 or S
When a negative determination is made in step 40 and the processing shifts to S50, data from each of the disk devices 31 to 34 exists, so the data of the disk devices 31 to 33 of A to C among them should be sent to the main computer 10. Data is obtained.

On the other hand, if the determination in S40 is affirmative,
Since there is a possibility that the data to be sent to the main computer 10 may not be obtained as it is, the process proceeds to a predetermined process after S60. First, in S60, it is determined whether or not there is one disk device that does not respond. If the number is one (S60: YES), it is determined at S70 whether the unresponsive disk device is the D disk device 34. If it is the D disk device 34 (S70: YES), the process directly proceeds to S50. Since the D disk device 34 stores the parity, the other disk devices 31 to 33 of A to C are used.
This is because if the data of 1 is obtained, there is no problem in the processing of S50. That is, data to be sent to the main computer 10 can be obtained from the data of the disk devices 31 to 33 of A to C.

However, a negative determination is made in S70, that is, if one disk device that does not respond is one of the disk devices 31 to 33 of A to C, the disk device is sent to the main computer 10 as it is. Data is not available. Therefore, the process proceeds to S80 and the corresponding data is created by error correction.

As described above with reference to FIG. 2, this is because the data is read from the remaining normal disk devices of the disk devices 31 to 33 for storing data which have no error. Based on one data and the parity data read from the D disk device 34, the data corresponding to the disk device in which the error has occurred is created (restored) by the error correction function.

After the data is created in S80, the process proceeds to S50. Since the data to be sent to the main computer 10 is obtained from the two data read from the normal disk device and the data created in S80, the data is sent to the main computer 10.

Incidentally, if a negative determination is made in S60, that is, if there are two or more disk devices that do not respond at the time-out, the error correction in S80 described above cannot deal with it. In that case, the process proceeds to S61. In S61,
Similar to S20, if data is transferred from each of the disk devices 31 to 34, it is stored in a buffer memory (not shown) in the disk array controller 35, and in subsequent S62, whether or not there is one or less disk device with no response. To judge. If the number is less than 1 (S6
2: YES), the process proceeds to S70, but if it is not one or less (that is, two or more) (S62: NO), S
It is judged whether or not a predetermined time has elapsed after shifting to 63 (whether or not it has timed out). If it has not timed out (S63: NO), the process returns to S61 to repeat the processing from S61 onward, but if it has timed out (S6
3: YES), shift to S90, main computer 1
A predetermined error notification is sent to 0, and this processing ends.

That is, when the responses from two or more disk devices are delayed, the number of disk devices that do not respond is 1
Wait for the end of the response until it is below the limit. This is because an error does not always occur even if the response is delayed due to the read retry, so that the response is waited for a predetermined time. It is sufficient that the number of disk devices that do not respond within the predetermined time is one or less. If not, an error notification will be sent in S90.

The timeout in S30 is set to a relatively short predetermined time, and the timeout in S63 is set to a relatively long predetermined time. In this way, when the processing described with reference to FIG. 3 is executed, the respective disk devices 31 to 34 are
When the normal reading operation is executed in, there is no difference from the conventional method, but in the situation where an error occurs, the improvement of the response is realized. That is, this is the case where the processing goes through the processing of S80 in FIG.

This point will be illustrated with reference to FIG. In response to a data read request from the main computer 10 (see FIG. 4A), the disk array controller 35
Issues a read command to the disk devices 31 to 34 of A to D (see FIG. 4A). FIG. 4B shows an example of a state after a predetermined time has elapsed from issuing the read command. In this case, the three disk devices 31, 32, 34 of A, B, D
, The data transfer is completed because a normal read operation has been performed, but the disk device 33 of C does not perform a normal read operation and, for example, the read retry process is being executed, so the data transfer is not completed yet. There is no response (see FIG. 4 (b)).

In such a situation, conventionally, until the operation of the C disk device 33 is completed and the C disk device 33 is notified that the data cannot be read due to an error, A, Data could not be created by the error correction function based on the data obtained from the B2 disk devices 31 and 32 and the parity data from the D disk device 34. That is, even if the operation is completed in the A and B disk devices 31 and 32 (even if data is transferred), C
The operation is terminated in the disk device 33, and it is necessary to know that there is an error there for the first time, so that the process cannot be shifted to the error correction process using the parity data of the D disk device 34. The response to the request from the main computer 10 will be considerably lowered.

On the other hand, the present disk array device 30
In this case, regardless of whether data is finally obtained from the disk device 31 to 34, for a disk device that does not respond at the time when a predetermined time has elapsed after the disk array controller 35 issued the read command, It is considered that a read error or the like is being executed due to the occurrence of a read error, and when there is only one device, the data is sent to the main computer 10 by the data from another disk device without waiting for the end of the operation. It creates and sends the correct data.

That is, as shown in FIG. 4B, even if there is no response from the C disk device 33 at the time-out,
As shown in FIG. 4C, two disk devices 3 of A and B are provided.
The data corresponding to the C disk device 33 in which the error has occurred is created (restored) by the error correction function based on the two data read from the first and the second data 32 and the parity data read from the D disk device 34. Then, the created data and the disk devices 31, A and B,
Since the data to be sent to the main computer 10 is obtained from the two data read from 32, the data is sent to the main computer 10.

When the operation of the D disk device 34 is delayed, the data read from only the disk devices 31 to 33 of A to C is sufficient, as described in the description of FIG. Only then, the data to be sent to the main computer 10 can be obtained.

Also, there was no response at the time of timeout 1
When data is sent to the main computer 10 without waiting for the operation of one disk device to be completed, the operation of one specific disk device is not completed at the next read request from the main computer 10. It is possible that it has not ended. In this case, since the read command itself cannot be issued to the disk device, the read command is output only to the remaining three units. And if there is data transfer from the three units,
On the basis of the three data, if necessary, error correction may be performed to create insufficient data, and the requested data may be sent to the main computer 10.

As described above, according to the disk array device 30, the capacity can be increased, the processing speed can be improved, and the error correction for the data error occurrence in the disk devices 31 to 34 can be performed, so that the reliability can be improved. The basic action and effect of being able to do is demonstrated.

In addition to them, when a read command is issued to the disk devices 31 to 34 in response to a request from the main computer 10, one of the disk devices 31 to 34 is read.
If the response from the stand does not elapse a predetermined time after issuing the command, do not wait for the response from the disk device,
Based on the data transferred from other disk devices,
If necessary, the error correction function can be used to create insufficient data and send the requested data to the main computer 10.

As a result, even if there is a disk device that takes more time to complete its operation than a normally completed disk device due to a read retry process being executed due to a read error, etc. Data to be transferred to the main computer 10 can be created by the error correction function by ignoring
The operation can be continued without affecting the overall operation speed.

That is, in the past, since the response was delayed from the disk device, the data was restored by error correction after it was determined that the necessary data could not be obtained due to an error at that time, so the disk array It will affect the operation speed of the entire device 30,
The response to the command from the main computer 10 was degraded. On the other hand, according to the disk array device 30 of the present invention, since the processing is performed without waiting for the response from the disk device that deteriorates such a response, the response of the entire disk array device is not deteriorated.

As described above, the present invention is not limited to such an embodiment, and can be implemented in various forms without departing from the gist of the present invention. In the case of the disk array device 30 shown in FIG. 1, A to C for data storage
3 disk units 31 to 33 and D for parity storage
Although the disk device 34 and the disk device 34 are provided as one set, two or more sets of such disk devices may be provided. For example, a disk array device according to another embodiment shown in FIG. 5 has three disk devices A1 to C1 for data storage.
01-103 and D1 disk device 10 for parity storage
4 and a second set including three disk devices 111 to 113 A2 to C2 for data storage and a D2 disk device 114 for parity storage, A3 to data storage. Three C3 disk devices 121 to 12
3 and a D3 disk device 124 for parity storage may be provided as a third set, .. In this case, if the disk controller shown in FIG. 5 executes the error correction by the parity check method for each set of the disk devices, it is more effective in realizing a large capacity and improving the processing speed. . It is also effective when multitask processing is requested from the connected main computer.

In the case of FIG. 5, a total of four data storage disk devices and one parity storage disk device make up one set, but in this case, for example, nine data storage disk devices. It is also conceivable to configure a total of 10 data storage disk devices and one parity storage disk device.
However, if the response is slow for any one of the 10 disk devices in total, data can be created by the error correction function by the parity check method. It becomes impossible to create the data about the stand by the error correction function. As a result, the entire configuration of 10 disk devices cannot be used, which poses a large risk. Therefore, for example, by configuring them with a total of four sets of three data storage disk devices and one parity storage disk device, the probability that the response of two of the four devices will be delayed is Since it is much smaller than the case of 2 units out of 10, it is effective in improving reliability.

In the above description, one of the error correction functions is based on the parity check method, but an error correction function of any other method may be used.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing a schematic configuration of a disk array device and a main computer connected thereto.

FIG. 2 is an explanatory diagram of error correction by a parity check method.

FIG. 3 is a flowchart showing a read control process executed in the disk array controller.

FIG. 4 is an explanatory diagram of processing when the end of operation of one of the disk devices in the disk array device is delayed.

FIG. 5 is an explanatory diagram showing another embodiment.

[Explanation of symbols]

10 ... Main computer 20 ... Bus 30 ... Disk array device 31 ... A disk device 32 ... B disk device 33 ... C disk device 34 ... D disk device 35 ... Disk array controller 101-104, 111-114, 121-124 ... Disk apparatus

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Office reference number FI Technical display location G11B 20/18 522 G11B 20/18 522D 570 570Z 572 572F

Claims (3)

[Claims] 1. A disk array controller is configured to control a plurality of disk devices and respond to an access from a main computer by masquerading as one disk device. In a disk array device having an error correction function for error occurrence, when the disk array controller issues a read command to a plurality of disk devices in response to a request from the main computer, among the plurality of disk devices If a response from a part of the above does not occur within a predetermined time after issuing the command, the data corresponding to the disk device which does not respond is created by the error correction function and requested by the main computer. Disk, which is characterized in that it is configured to send Array device. 2. When the disk array controller issues a read command to a plurality of disk devices in response to a request from the main computer, the current read command is issued because the operation for the previous read command has not been completed yet. When there is a disk device that cannot be issued, data corresponding to the disk device for which the read command cannot be issued is created by the error correction function, and the requested data is sent to the main computer. The disk array device according to claim 1, wherein: 3. A predetermined number of data storage disk devices operating in parallel, and a parity in which a parity created for each parallel data recorded in a predetermined same storage position of each data storage disk device is recorded. A plurality of sets of disk devices each including a storage disk device are provided, and the disk controller is capable of performing error correction by a parity check method for each set of the disk devices. 1. The disk array device according to 1 or 2.