JPH10171608A - Magnetic disk array device and method for checking data in magnetic disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data checking method capable of specifying a magnetic disk device which is not detected by data check (CRC, parity) to be executed in each magnetic disk device in a disk array device and causes a logical fault. SOLUTION: In the magnetic disk array device 20, a control means 7 is provided with an input data checking means 17 for checking data by using a check code and information adding means 15, 16 for adding address information and information for expressing a time element to data and each of magnetic drive devices 1 to 4 includes a received signal analyzing means 12 for analyzing a control signal received from a host means 5, a received data checking means 13 for checking data received from the host means 5 and a data storage means 14.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a magnetic disk array device and a data check method in the magnetic disk array device, and more particularly to a method for controlling data storage means of the magnetic disk array device. .

[0002]

2. Description of the Related Art Conventionally, generally referred to as a plurality of physical drives or devices, magnetic disk devices are generally arranged in parallel, and commands from a host means such as a computer, which is a higher-level device, are transmitted. An array disk device in which a plurality of devices (magnetic disk devices) are operated in parallel by a command or the like to simultaneously input and output data has been used in many fields.

Such a magnetic disk array device is an excellent system in that data can be efficiently processed and high-speed data transfer can be performed. However, in such a magnetic disk array device, for example, in a data path control circuit, a complicated data transfer path switching operation is required, so that a failure occurs on the data transfer path or in the magnetic disk device. Data corruption will occur.

In such a system, usually, both a parity check and a CRC check are performed to constantly check whether or not the transferred data is correct. However, for example, a command for reading predetermined data stored at the same address from each of the plurality of magnetic disk devices is input from the host means side to the magnetic disk device side, and a predetermined address is read from each magnetic disk device. When one of the plurality of magnetic disk devices erroneously reads data at a different address when reading the data stored in the Although an error is detected in the logical parity, the CRC in the data read from each magnetic disk device is
Are correct, so the CRC check is not caught.

In other words, there is a problem that it is not possible to detect from which address of which magnetic disk device the data read is erroneous. If such a state occurs frequently, the reliability of the data is reduced and the reliability of the magnetic disk array device is also reduced. In this case, in the worst case, it is necessary to replace the magnetic disk device itself.

In order to improve such a problem, for example, Japanese Patent Application Laid-Open
Japanese Patent Application Publication No. 166304 discloses a data check method for a magnetic disk array device having a circuit configuration as shown in FIG. That is, a plurality of devices (DV0,
DV1, DV2 ... DV7, DVP, DVH), for example, an array that has a physical drive such as a magnetic disk drive, and controls such magnetic disk drives to operate in parallel and to simultaneously input and output data. In the data check method for a disk device, when a write data block including data for CRC (cyclic code inspection) created on the host device side and written to a magnetic disk array device or the like is received, the CRC data is checked. The CRC data is stored in the internal register 18 and the data block is divided for simultaneous writing to a plurality of magnetic disk devices.
For each divided data, create CRC data for the device and write it to the device, and at the same time transfer the write data to the read data transfer path,
The data to be sent to the host means is combined, the CRC data of the host means is created again for each combined data, and the CRC data stored in the register 18 is stored in the register 18.
A data check method for checking write data to the magnetic disk device by checking the data in comparison with the C data is disclosed. Also, a method for checking a CRC read from a device such as a magnetic disk device is disclosed. When a data block including data is received, the CRC data is checked, the CRC data is stored in the internal register 19, and data to be sent from the data block to the host means is synthesized. Create CRC data for the host means for each data,
Simultaneously with the transfer to the host means, the transfer data to the host means is transferred to the write data transfer path, and the data to be written to the device is divided. Also, a method of checking data read from a device by creating the above data and comparing and checking the CRC data with the CRC data stored in the register 19 is disclosed.

In FIG. 2, 21 is a write data conversion unit, 20 is a CRC check unit on the host means side, and 25 is
A CRC generator on the host means side, a read data converter 26, a CRC check circuit 24, and multiplexers 22 and 23, respectively.

[0008]

However, in the above-mentioned prior art, the first problem is that the CRs are transmitted through different transfer paths.
Since C is generated, an actual data transfer roller and a data transfer path for data check are different. Therefore, if there is an abnormality in the data transfer path, it cannot be determined which CRC is correct.

A second problem is that if a certain device can read data (block address different from the block requested by the higher-level device) different from the data requested by the higher-level device due to data corruption or the like, no abnormality can be detected. The reason is that in the CRC check for each device, since the data of the correct address has been read out temporarily, the error cannot be detected, and furthermore, the CRC sent to the host means side during the read operation is from the device. CR from received data
This is because C is being remade.

[0010] Therefore, an object of the present invention is to reconstruct data as logic without being detected by data check (CRC, parity) performed for each physical device, that is, for each magnetic disk device in the disk array device. The present invention provides a magnetic disk array device that can identify a physical device that causes a failure in a logical failure that occurs in the event that the magnetic disk device is used, and uses the magnetic disk device. A data check method is provided.

Further, the present invention provides a magnetic disk array device which particularly contributes to improvement in reliability.

[0012]

SUMMARY OF THE INVENTION The present invention basically employs the following technical configuration to achieve the above object. That is, as a first aspect of the present invention, there is provided a magnetic disk array device configured so that a host means drives a plurality of magnetic drive apparatuses in parallel via a control means to execute data input / output. In the control means, there are provided input data checking means for confirming data using a check code, and information adding means for adding address information and information representing a temporal element to the data. Each of the magnetic drive devices includes a reception signal analysis unit that analyzes a control signal from the host unit, a reception data check unit that checks data received from the host unit, and a data storage unit. In a second mode, the host device drives a plurality of magnetic drive devices in parallel via the control device to perform data decoding. Confirming predetermined data transferred from the host means using a check code in the magnetic disk array device configured to execute input / output of data, When the data is divided into data groups and transmitted to each of the plurality of magnetic drive devices,
A step of adding at least address information and information representing a time element to each of the divided data, and responding to the write command from the host means to store the divided data in a predetermined format of each magnetic drive device. A data check method in the magnetic disk array device comprising the step of storing data in the storage means.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The magnetic disk array device and the data check method in the magnetic disk array device according to the present invention employ the above-mentioned technical configuration. Since the data is checked by the array controller using the check code for each data, in the conventional magnetic disk array device, the data corruption failure which has been processed as a logical process,
It can be handled as a physical drive failure.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A specific example of a magnetic disk array device and a data check method in the magnetic disk array device according to the present invention will be described below in detail with reference to FIG. FIG. 1 is a block diagram showing a configuration of a specific example of a magnetic disk array device according to the present invention. In the drawing, a host means 5 controls a plurality of magnetic drive apparatuses 1 to 4 in parallel via a control means 7. In the magnetic disk array device 20 which is configured to execute data input / output by driving the data, the control means 7 includes an input data check means 17 for checking data using a check code, Information addition means 15 and 16 for adding address information and information representing a time element are provided, and each of the magnetic drive devices 1 to 4 receives and analyzes a control signal from the host means 5. A magnetic disk including a signal analysis unit 12, a reception data check unit 13 for checking data received from the host unit 5, and a data storage unit 14. Array device 20 is shown.

The magnetic disk array device 20 according to the present invention
In FIG. 1, the magnetic disk devices 1 to 4 as four physical drives are used in FIG. 1, but the present invention is not limited to the number, and the number is arbitrary. It can be increased or decreased as needed. However,
Of the plurality of magnetic disk devices in the magnetic disk array device 20 according to the present invention, one magnetic disk device, for example, the magnetic disk device 4 functions as a parity drive.

In the present invention, the plurality of magnetic disk devices 1 to 4 are connected to control means 7 called an array controller via appropriate interfaces 8 to 11, respectively. The means 7 is connected to a host computer 5 serving as a host means via an appropriate interface 6 so that predetermined data can be transferred to the magnetic disk devices 1 to 4 via the control means 7.
And the host means 5 are transferred in both directions.

The input data checking means 17 in the control means 7 according to the present invention comprises: data transferred from the host means 5 to the plurality of magnetic drive devices 1-4; It is preferable to have a function of checking both data transferred from the drive devices 1 to 4 to the host means 5 side. The input data check means 17 used in the present invention, for example, performs a total CRC (TOTAL CR) on each of the divided data.
C) or has a function of setting a logical parity for the combined data.

Further, the control means 7 in the present invention comprises:
From the host means 5 side, a plurality of magnetic drive devices 1 to
When the data is transferred to the magnetic disk devices 1 to 4, the data is divided and transferred to the respective magnetic drive devices 1 to 4, or the data read from each of the plurality of magnetic disk devices 1 to 4 is synthesized. It has a data transfer means 18 having a function of transferring to the host means 5.

In the present invention, as the storage means 14 provided in each of the plurality of magnetic drive devices 1 to 4, one having an appropriate storage function including a hard disk can be used. A register unit for individually storing at least address information, information representing a temporal element, data, a CRC code, and a total CRC code of data divided by the control means 7 and transferred from the control means 7; Is desirable.

The magnetic disk array device according to the present invention may be constituted by a system generally called RAID3. In this case, the magnetic disk array device is constituted by four magnetic disk devices 1 to 4 of one logic. If one of the magnetic disk units fails, the logic of the remaining three units is degenerated. In the magnetic disk array device 20 according to the present invention, when the host unit 5 issues a write command, the host unit 5 transmits the write destination block address and data to the array controller 7 which is the control unit. It is configured to perform an operation of adding a write destination block address together with data to the array controller 7, and to add the time at which the host unit 5 issues a write command to the array controller 7 and send the data. ing.

Therefore, in the present invention, the control means which has received the write command, that is, the array controller 7 checks whether or not the data received from the host means 5 by the input data check means 17 is correctly received.
A check (CRC check) is performed using a code, data is divided and developed so as to correspond to the array configuration, and is sent to each of the magnetic disk devices 1 to 4 constituting each physical drive.

For example, at the beginning of a block address of data sent to each of the magnetic disk devices 1 to 4, time information at which a write command is issued from the host means 5 and block address information of a write destination are added. The control means 7
Magnetic disk drives 1 that have received data from
When the command decode circuit 12 determines that the instruction is a write operation, the check circuit 13 checks the data received from the array controller 7 (CRC check). If the data is received correctly, Then, writing is performed on each storage means 14, for example, a hard disk (HDD).

An example of a data format used in the present invention and an example of a storage format of the storage means in each of the magnetic disk devices 1 to 4 will be described with reference to FIG. That is, FIG. 3A is a diagram showing an example of the structure of the data transferred from the host means 5 to the control means 7 in the present invention. ADD), and thereafter, data 1 (D) which is individually distributed to each of the magnetic disk devices 1 to 4
AT1), data 2 (DAT2), data 3 (DAT
3)... And the following sections contain data (DAT1), data 2 (DAT2), and data 3 (DA
T3) check information (CRC1,
CRC2, CRC3...) Are arranged.

On the other hand, when the data shown in FIG. 3A is stored in one storage means 14 of each of the magnetic disk devices 1 to 4, the storage format is as shown in FIG. , Information representing a temporal element, block address information, data (DAT1 to DATN), CRC (1 to n) information corresponding to each of the data, and a total CRC in each of the magnetic disk devices 1 to 4. It is configured to include.

Next, when a read command is issued from the host means 5 to the control means 7, the control means 7 issues a read command to each of the magnetic disk devices 1-4.
Each of the magnetic disk devices 1 to 4
4 from the check circuit 13 (CR
Check in (C Check). At this time, the magnetic disk devices 1 to 4 that have detected a read error or a CRC error
Then, the controller 7 is notified of the content of the fault.

The control means 7 receiving the contents of the fault
Performs a retry on the failed magnetic disk devices 1 to 4 and, if the failure is not recovered, degenerates the magnetic disk device and continues operation of the magnetic disk array device in a degraded state. I do. Next, when it is determined that the data is normal in each of the magnetic disk devices 1 to 4, the data is sent to the control unit 7, and the control unit 7 sends the data to each of the magnetic disk devices 1 to 4. First, the input data is checked by the input data check means 17 for CR.
Perform C check.

If the CRC check is normal, a check of a time stamp used as information including a block address value and a time element included in data transmitted from each of the magnetic disk devices 1 to 4 is executed. As a specific example of the method, first, a time stamp relating to data stored in the storage unit 14 of each magnetic disk device is checked, and majority processing is performed.

For example, when only one of the four magnetic disk devices has a different time stamp, the one magnetic disk device is degenerated and the magnetic disk array device is operated in a degraded state. Next, a comparison between the block address value included in the data and the block address value requested by the host unit 5 is performed for all magnetic disk devices for each block unit of each magnetic disk device.

Next, if the block address value requested by the host means 5 matches the block address value of the data downloaded from each magnetic disk device, the means is transferred to the host means 5. As described above, as a data check method of a magnetic disk array device according to the present invention, as a basic configuration, a host device drives a plurality of magnetic drive devices in parallel via a control device to input data. A step of confirming, using a check code, predetermined data transferred from the host means in the magnetic disk array device configured to execute the output, and converting the predetermined data into a plurality of data groups. Dividing and transmitting, to each of the plurality of magnetic drive devices, at least address information and information representing a time element for each of the divided data;
And in response to a write command from the host means,
Storing the divided data in a predetermined storage means of each magnetic drive device. Preferably, the divided data is transferred from the host means in each of the plurality of magnetic drive devices. The method further comprises a step of checking the divided data, creating a check code, and storing the check code in the storage means.

Further, as a data check method for a magnetic disk array device according to the present invention, in addition to the above-mentioned configuration, a step of analyzing a command from a host means in each of the plurality of magnetic drive devices, When the command is a read command, a step of reading predetermined data from a predetermined address portion from a storage unit of the magnetic drive device, and reading the data read from the storage unit of each magnetic drive device with the address information. And a step of checking for errors using information representing a temporal element.

[0031]

As described above, the magnetic disk array device and the data check method of the magnetic disk array device according to the present invention adopt the above-described configuration, and therefore, the storage means of each magnetic disk device can be used. Even if old data remaining in a certain cache register is read by mistake, a part of that data, such as a time stamp, is converted to data for each magnetic disk device that composes each physical drive. Since information indicating a target element is additionally stored, there is an effect that an abnormality in data can be detected when reading or writing data.

In the present invention, even when data is written to a block address different from the block address to which data is to be written, or when data is read from a block address different from the block address to be read, each magnetic disk drive Since a block address value as address information is added to a part of the data at the time of conversion to data, there is an effect that data abnormality can be detected.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of one specific example of a magnetic disk array device according to the present invention.

FIG. 2 is a block diagram showing a configuration example of control means in a conventional magnetic disk array device.

FIG. 3A is a data format showing an example of a data structure used in the present invention.
FIG. 3B is a diagram showing an example of a configuration of a register unit of a storage unit of each magnetic disk device according to the present invention.

[Explanation of symbols]

 1-4 physical drive, magnetic disk drive 5 host means, host computer 6, 8-11 interface 7 control means, array controller 15 block address distribution circuit 16 time stamp distribution circuit 17 input data check circuit 18 ... Data distribution and synthesis means 20 ... Magnetic disk array device

Claims (9)

[Claims] 1. A magnetic disk array device wherein a host means drives a plurality of magnetic drive devices in parallel via a control means to execute data input / output. Is provided with input data checking means for confirming data using a check code, and information adding means for adding address information and information representing a time element to the data, and each of the magnetic drive devices is provided with A magnetic disk array device comprising: a received signal analyzing means for analyzing a control signal from the host means; a received data checking means for checking data received from the host means; and a data storage means. 2. The input data check means in the control means, wherein data transferred from the host means to a plurality of magnetic drive devices and data transferred from the plurality of magnetic drive devices to the host means side. 2. The magnetic disk array device according to claim 1, wherein both of the data are checked. 3. The control means has data transfer means for, when data is transferred from the host means to a plurality of magnetic drive devices, dividing the data and transferring the divided data to the respective magnetic drive devices. 3. The magnetic disk array device according to claim 1, wherein: 4. The storage means provided in each of the plurality of magnetic drive devices stores at least address information and a time element included in data divided by the control means and transferred from the control means. 4. The magnetic disk array device according to claim 1, further comprising a register unit for individually storing information, data, a CRC code, and a comprehensive CRC code. 5. The input data check means in the control means detects a data error by referring to address information and information representing a time element included in data output from each magnetic drive device. 5. The magnetic disk array device according to claim 1, wherein the magnetic disk array device is configured as follows. 6. The storage means in one magnetic drive device of the plurality of magnetic drive devices stores parity relating to respective information stored in a register unit in storage means of another magnetic drive device. 6. The magnetic disk array device according to claim 1, further comprising a register unit for performing the operation. 7. A magnetic disk array device in which a host means drives a plurality of magnetic drive devices in parallel via a control means to execute data input / output. A step of confirming the transferred predetermined data using a check code, dividing the predetermined data into a plurality of data groups, and transmitting the divided data to each of the plurality of magnetic drive devices. A step of adding at least address information and information indicating a time element to each of the divided data; and, in response to a write command from the host means, storing the divided data in a predetermined storage means of each magnetic drive device. A data check method in the magnetic disk array device. 8. In each of the plurality of magnetic drive devices, the method further comprises a step of checking the divided data transferred from the host means, creating a check code, and storing the check code in the storage means. 8. The data check method according to claim 7, wherein: 9. A step of analyzing a command from a host means in each of the plurality of magnetic drive devices, and, if the command is a read command, a predetermined address from a storage means of the magnetic drive device. Reading predetermined data from a part, and checking the data read from the storage means of each magnetic drive device for errors using the address information and information representing a time element. 9. The data check method according to claim 7, wherein: