JP2001142650A - Method and device for controlling array disk - Google Patents

Abstract

PROBLEM TO BE SOLVED: To find the occurrence of illegal data and a disk failure at an early stage and to reliably store even data that are not read for a long time. SOLUTION: This device is provided with a cache memory 6, a data dividing part 3 which divides data transmitted from a host device 1 in accordance with the number of physical disks and also generates the parity data of the divided division data, a data transferring part 4 which respectively stores the division data and the parity data into physical disks, a data comparing part 8 which reads the data from the physical disk after storing the division data and the parity data in each physical disk and compares the data with corresponding data stored in the cache memory, and a failure deciding part 10 which decides that a failure takes place in the physical disk storing an uncoincidental part when any uncoincidental part exists in the part 8.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an array disk control method and apparatus, and more particularly to an array disk control method and apparatus for dividing data into a plurality of physical disks and storing the data with redundancy.

[0002]

2. Description of the Related Art Conventionally, data transferred from a host device is divided, parity data of the divided data is generated, and these are stored in separate physical disks.
A technique that can recover data even when a failure occurs in one physical disk is used. For such data division and generation of parity data, for example, R
AID3, 5, etc.

However, in the above-mentioned conventional example, a failure has occurred in the physical disk, and data can be read. However, if the data content is incorrect, it is possible to determine which physical disk has failed. Can not. On the other hand, in the method described in JP-A-9-305328,
In addition to generating parity data from the divided data,
Cyclic redundancy check information (CRC) is added. And
If the parity of the reproduced data is incorrect, 1
The data is restored by a plurality of combinations in the case where the unit is a failed disk, and among the plurality of restored data, the restored data determined to be normal based on the CRC is transferred to the host device as reproduction data.

[0004]

However, in the above-mentioned conventional example, since the illegal data is checked at the time of reading the data, if the data is not read for a long time after the data is written, the cause of the illegal data is clarified. If it is difficult to identify it, then if a long time has elapsed after data writing, it is assumed that failures will occur on multiple physical disks at the same time, and it is highly likely that data recovery will be difficult , There was an inconvenience.

[0005]

SUMMARY OF THE INVENTION The object of the present invention is to improve the disadvantages of the prior art and, in particular, to detect the occurrence of illegal data and disk failure at an early stage, and to reliably store even data that cannot be read for a long time. An object of the present invention is to provide a method and apparatus for controlling an array disk.

[0006]

Therefore, according to the present invention, there are provided a plurality of physical disks constituting a logical disk, a cache memory for temporarily storing data transmitted from a higher-level device, and a cache memory for storing data transmitted from the higher-level device. A data division unit that divides the divided data according to the number of the physical disks and generates parity data of the divided data, and stores the divided data and parity data generated by the data division unit in each of the physical disks. A disk array control method for storing data transmitted from a higher-level device using a disk array device having a data transfer unit for storing data before and after storing divided data and parity data in each of the physical disks. And transmitting a storage completion report command to the host device. A data comparison that reads the data from each of the physical disks and compares the read data with the corresponding data stored in the cache memory until the data for which the storage completion report has been made is deleted from the cache memory after the reporting step. And a failure determination step of determining that a failure has occurred in the physical disk that stores the mismatched portion if there is a mismatched portion in the data comparison process. This aims to achieve the above-mentioned object.

In the present invention, data transmitted from a host device is first stored in a cache memory. When the data is stored in the cache memory, a storage completion report command is transmitted to the host device. Then, while dividing the data stored in the cache memory, parity data is generated and stored in a plurality of physical disks. The storage completion report to the host device may be made when the storage of the divided data and the parity data on the physical disk is completed.

Subsequently, when there is another I / O request from the host device, the I / O request is processed. Then, before deleting the same data as the data stored in each of the physical disks from the cache memory, it is determined whether the storage of the data in the physical disks has been completed normally.
That is, in the data comparison step, the data is read from each physical disk and compared with the corresponding data stored in the cache memory. Then, in the failure determination step, when there is a part that does not match in the data comparison step, it is determined that a failure has occurred in the physical disk storing the part that does not match. As described above, by directly comparing the data stored in the cache memory with the data stored in each physical disk, it is confirmed whether or not illegal data has occurred for each physical disk.

[0009]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of an array disk control device according to the present invention. As shown in FIG. 1, the array disk control device 2 includes a cache memory 6 for temporarily storing data transmitted from the higher-level device 1, and stores the data transmitted from the higher-level device 1 in accordance with the number of the physical disks. Data dividing unit 3 that divides the divided data and generates parity data of the divided data.
And a data transfer unit 4 for storing the divided data and parity data generated by the data dividing unit in each of the physical disks. Data division unit 3
May be a function of the data transfer unit 4 without being provided in the data transfer unit 3.

After storing the divided data and the parity data in each physical disk, the array disk controller 2 reads the data from each of the physical disks and compares it with the corresponding data stored in the cache memory. The data comparison unit 8 includes a failure determination unit 10 that determines that a failure has occurred in a physical disk that stores the mismatched portion when there is a mismatched portion in the data comparison unit 8.

The array disk control device is connected to the host device 1 and each of the physical disks 21, 22, 23 constituting the logical disk 20 via the data transfer unit 4. Here, N physical disks from physical disk 1 to physical disk N are provided.

The disk transfer unit 4 transfers data to and from the host device 1, reads / writes data to / from the cache memory 6, and transfers data to the logical disk 20. The cache memory uses a storage medium, such as a RAM, which is faster to access than a physical disk.

The data comparing section 8 reads the data stored in each physical disk in a state where the write data received from the host device is stored in the cache memory 6 and compares the data with the data stored in the cache memory 6. . At this time, if there is a mismatched portion, it means that writing has not been performed normally. Then, it is determined based on the division rule by the data division unit 3 which physical disk the data of the part that has not been normally written is the data that has been written. For this reason, if there is a mismatch in the compared data, it can be determined on which physical disk the incorrect data that caused the mismatch was stored.

Accordingly, when there is a part that does not match in the data comparing unit 8, the failure determining unit 10 determines that a failure has occurred in the physical disk storing the part that does not match.
As a result, it is possible to specify a physical disk in which a failure has occurred such that data can be read but writing cannot be performed normally, and recovery processing can be performed. Then
Even if data is not read for a long period of time, it is possible to check whether the data was stored correctly or not, for example, to prevent a situation in which multiple physical disks failed at the same time and the data could not be recovered can do. Also, since the physical disk that could not be written normally during data storage is specified, the environment in which illegal data has occurred is maintained by notifying the host device of the physical disk device number and the like as write failure information. The cause of the occurrence of the incorrect data can be clarified as it is, and therefore, the cause of the failure can be more easily clarified as compared with the case where the incorrect data occurs after a long period of time.

When the failure determining unit 10 identifies the failed physical disk, the recovery processing unit 12 reads the divided data including the invalid data portion from the cache memory 6 and performs the rewriting process once or It is repeated several times, and it is again compared with the data stored in the cache memory to determine whether or not illegal data has occurred. At this time, writing may be performed to a different sector of the physical disk in which the illegal data has occurred. If illegal data still occurs after such rewriting,
Degenerate the physical disk that caused the illegal data. The degeneration process is a process of restoring all other data and then disconnecting the physical disk.

FIG. 2 is an explanatory diagram showing an example of division of data transmitted as write data from a host device (host processor). In the cache memory 6, FIG.
The data received from the host processor as shown in FIG. On the other hand, the data to be recorded on the logical disk is obtained by dividing the data received from the host processor data by the data dividing unit 3 and rewriting each physical disk 1 (reference numeral 2).
1), physical disk 2 (reference numeral 22), physical disk N
(Reference numeral 23). When data is recorded on a logical disk 20 composed of N physical disks, FIG.
As shown in (B), the data is divided into N-1 pieces and divided data 1 (code 31), divided data 2 (code 32) to divided data N-2 (code 33), and divided data N-1 (code 31). Reference numeral 34) is generated. Then, parity data 35 of the divided data is generated. As shown in FIG. 2C, each of the physical disks 1 to N has divided data 1 to N-1.
Then, the generated parity data 35 is written. When reading the data, all the divided data 1 to N-1 and the parity data 35 are read, a parity check is performed to confirm the validity of the data, and the data excluding the parity data 35 is transferred to the host processor 1.

Next, an example of an operation of controlling an array disk using the array disk control device shown in FIG. 1 will be described with reference to FIG. FIG. 3 is a flowchart illustrating an operation example of the present embodiment. First, the data transfer unit 4 receives write data from the host device (host processor) 1 (step S1), and stores the data in the cache memory 6 (step S2). Further, the data dividing unit 3
Generates the parity data 35 by dividing the data (step S3). The data transfer unit 4 stores the divided data 31 to 34 and the parity data 35 in each of the physical disks 21 to 23 (Step S4). continue,
The data transfer unit 4 reports the storage completion to the host processor 1 (step S5, storage completion reporting step). Specifically, a command indicating a storage completion report is transmitted to the input / output interface of the higher-level device. The storage completion report to the host device may be made immediately after the storage in the cache memory 6 in step S2 is completed, not immediately after the data is actually stored in each physical disk.

In the example shown in FIG. 3, following the storage completion report S5, that is, after disconnecting the host device, the data stored in the physical disk is compared with the data stored in the cache memory 6 (data Comparison steps S6 to S
8). The data transfer unit 4 reads the data stored in the physical disk (Step S7). On the other hand, the data comparison unit 8
Reads the same data from the cache memory 6 and generates parity data (step S
8). Then, the data comparing unit 8 compares the data read from the physical disk with the data stored in the cache memory 6 (Step S9). In the example shown in FIG. 3, the data comparison step includes a step of generating parity data of data read from the cache memory. This makes it possible to determine whether or not the data storing the parity data is invalid.

As a result of comparison between the two data, if they match, the writing has been performed normally, so the data writing process is terminated. On the other hand, if there is a difference between the two data,
It is determined that a failure has occurred in the physical disk storing the inconsistent portion (step S10, failure determination step).
On the other hand, when there is a difference between the two data, the recovery processing unit 12 is activated to perform the recovery processing (step S1).
1, restoration processing step).

The recovery processing step S11 includes a rewriting step of rewriting data stored in the disk cache to the physical disk determined to be the failure, and a degeneration of the physical disk determined to be the failure and the physical disk It is preferable to provide a degeneration step for disabling the use of.

In the example shown in FIG. 3, the data comparison step is positioned immediately after the storage completion report S5 to the host device 1 is completed and immediately after the data is stored in the physical disk. A data write request and a data reproduction request (I /
If (O request) is continuous, it is preferable to execute the data comparison step immediately after writing the data as shown in FIG. On the other hand, if the I / O request of the host device is intermittent,
The I / O request may be preferentially executed, and the data comparison step may be performed using the waiting time of the I / O request. In this case, in this embodiment, since the data stored in the cache memory 6 is used, the comparison of the data must be completed before the data stored in the cache memory 6 is deleted.

FIG. 4 is a flowchart showing an example of processing for comparing data immediately before data is deleted from the cache memory 6. In the example shown in FIG. 4, following the storage completion reporting step S5, a deletion target data selection step S24 for selecting data with a low access frequency from the host device among the data stored in the cache memory, A pre-deletion comparison step S27 for executing the data comparison step S9 on the data selected in the selection step S24 and deleting the data from the cache memory when the failure determination step S10 determines that there is no failure. ing.

Specifically, first, the I / O from the host device
The presence or absence of the request is confirmed (step S21). If the I / O request is a write, execution is not possible unless the cache memory 6 has a sufficient capacity, so it is confirmed whether or not the cache capacity is sufficient (step S22). If the cache capacity is sufficient, an I / O process is executed (step S23). That is, if data writing is requested, steps S1 to S5 shown in FIG. 3 are executed, and if data reading is requested, if there is corresponding data in the cache memory 6, the data is sent to the host device. On the other hand, if there is no corresponding data in the cache memory 6, the data is read from the physical disk, restored, and transferred to the host device. At the time of data read request,
When the data is stored in the cache memory 6, it is preferable to increase the value indicating the access frequency of the data.

When waiting for an I / O request from a higher-level device or when the cache capacity is insufficient, first, the data stored in the cache memory 6 is searched for data having a low access frequency. (Step S24).
Since this data search / identification is the identification of data to be deleted, other methods common in the field of array disks, such as selecting data after a certain period of time, can be adopted in addition to access frequency. .

When data to be deleted is specified, the data is compared with the data stored in each disk (step S25). That is, steps S6 to S9 shown in FIG. 3 are executed. If the two data match, the data is stored normally, and the cache data stored in the cache memory is deleted (step S27). If there is a sufficient free space in the cache memory, the fact that the cache memory can be deleted may be notified to a cache memory management unit (not shown) that manages the cache memory.

On the other hand, if the data does not match,
The recovery process S11 is executed.

As described above, according to the present embodiment, by comparing the data on the cache with the data written on the disk, it is determined whether or not the write data from the host device has been correctly written on all of the plurality of physical disks. Then, it becomes possible to prevent write-through of the divided data beforehand.

[0029]

Since the present invention is constructed and functions as described above, according to this, when a failure determination step includes a part that does not match in the data comparison step, a failure occurs in the physical disk storing the mismatched part. Is determined, the presence or absence of occurrence of illegal data can be confirmed for each physical disk immediately after the data is written. Then, the cause of the physical disk failure can be easily clarified and the data can be transmitted for a long time. The physical disk on which the illegal data was stored because the illegal data was identified by directly comparing the data stored in the cache memory with the data stored on the physical disk. Can be immediately identified which physical disk is, which increases the options for disaster recovery processing, for example,
An unprecedented superior array disk that enables early detection and recovery of a physical disk failure, such as performing degeneration processing on the physical disk if data corruption still persists after multiple rewrites. A control method can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first embodiment of the present invention.

2A and 2B are explanatory diagrams showing the structure of data handled by each component shown in FIG. 1. FIG. 2A is a diagram showing the structure of data transferred from a higher-level device, and FIG. FIG. 2 is a diagram showing a structure of data generated by a disk dividing unit;
(C) is a diagram showing a structure of data stored in each physical disk.

FIG. 3 is a flowchart illustrating an example of a data write process in the configuration illustrated in FIG. 1;

FIG. 4 is a flowchart illustrating an example of another data writing process in the configuration illustrated in FIG. 1;

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 High-order apparatus (host processor) 2 Array disk control apparatus 3 Data division part 4 Data transfer part 6 Cache memory 8 Data comparison part 10 Failure judgment part 12 Recovery processing part 20 Logical disk 21, 22, 23 Physical disk which comprises a logical disk

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G11B 20/18 501 G11B 20/18 501B 522 522Z 570 570Z 572 572B 572F

Claims (5)

[Claims] 1. A plurality of physical disks constituting a logical disk, a cache memory for temporarily storing data transmitted from a higher-level device, and data transmitted from the higher-level device according to the number of the physical disks. A data dividing unit for generating parity data of the divided data, and a data transfer unit for storing the divided data and parity data generated by the data dividing unit in each of the physical disks. A disk array control method for storing data transmitted from a higher-level device using an array device, comprising: a storage completion report to the higher-level device before and after a process of storing the divided data and parity data in each of the physical disks A storage completion reporting step of transmitting a command, and after the storage completion reporting step, the cache A data comparing step of reading the data from each of the physical disks and comparing the data with the corresponding data stored in the cache memory until the data for which the storage completion report has been made is deleted from the storage device; A failure determination step of determining that a failure has occurred in a physical disk that stores the mismatched portion when there is a mismatched portion in the process. 2. The method according to claim 1, further comprising a rewriting step of rewriting data stored in the disk cache to the physical disk determined to be the failure, following the failure determination step. Array disk control method. 3. A degrading step following the failure judging step, wherein the physical disk determined to be the fault is degraded to disable use of the physical disk. Array disk control method. 4. A deletion target data selection step of selecting data with a low access frequency from a host device among the data stored in the cache memory following the storage completion report step, and a deletion target data selection step. And a pre-deletion comparison step of causing the data comparison step to be performed on the selected data and deleting the data from the cache memory when the failure determination step determines that there is no failure. Item 2. An array disk control method according to Item 1. 5. A cache memory for temporarily storing data transmitted from a higher-level device, a cache memory for dividing data transmitted from the higher-level device in accordance with the number of physical disks, and parity data of the divided data. And a data transfer unit that stores the divided data and parity data generated by the data dividing unit in each of the physical disks, and stores the divided data and parity data in each of the physical disks. After that, a data comparison unit that reads the data from each of the physical disks and compares the read data with the corresponding data stored in the cache memory. A failure determination unit that determines that a failure has occurred in the stored physical disk Array disk control apparatus according to claim and.