JP3597766B2 - Disk array device control method - Google Patents

Description

[0001]
[Industrial applications]
The present invention relates to a disk array device called a so-called RAID (Redundant Array of Inexpensive Disks), and relates to an access processing performance of a disk device, in particular, to speeding up a data writing process.
[0002]
[Prior art]
Among the components constituting a computer, a hard disk device having a drive mechanism (also called a disk device or a drive) has a higher failure rate than other electronic components, and is used as a means for writing data, so that the failure causes The impact is great. Therefore, improvement in reliability has been demanded. Japanese Patent Application Laid-Open No. 2-236714 discloses a device that meets such a demand. It has a plurality of disk devices and, when writing storage data to be written, generates and writes redundant data in addition to writing the storage data.
[0003]
When one disk device fails and the data stored on the disk device cannot be read, the redundant data is the data that cannot be read from the redundant data and the data stored on the other failed disk device. Generated to be reproducible. The generated redundant data is written to another disk device different from the disk device storing the storage data used to generate the redundant data.
[0004]
A device composed of a plurality of disk devices configured in this manner is called a disk array device or RAID. According to this, even if any disk device constituting the array fails, the stored data on the failed disk device can be restored from the stored data of the other disk device and the redundant data, and the data integrity of the hard disk device can be restored. Improvement can be achieved.
[0005]
[Problems to be solved by the invention]
In RAID, although data integrity is improved, writing of stored data always involves generation and writing of the above-mentioned redundant data, which increases the write processing time and delays the external reporting of write completion. Therefore, the problem that the performance is reduced has been pointed out.
[0006]
Therefore, it is practically only suitable for a system in which writing to the disk device is small (read processing accounts for a large proportion).
[0007]
As a countermeasure against this, a cache memory is prepared in the disk controller, and when the writing to the cache memory is completed, the completion of writing is immediately reported to the host computer, and then the redundant data is generated. Some systems write to.
[0008]
However, this system has a problem that the system becomes expensive because an expensive cache memory is used.
[0009]
SUMMARY OF THE INVENTION An object of the present invention is to provide a disk array device which prevents the performance degradation (write penalty) during the write processing without increasing the cost.
[0010]
Another object of the present invention is to provide a disk array device that can maintain data integrity, which is a feature of RAID, in addition to the above objects.
[0011]
[Means for Solving the Problems]
In order to achieve the above object, a plurality of disk devices are provided, and at least one of the plurality of disk devices cannot read stored data written to another disk device when the data cannot be read. In a disk array device having a redundant data area for writing redundant data for reproducing data, in at least one of the disk devices, the area for writing the storage data to be stored temporarily writes the storage data. And a storage data area for finally writing the storage data written in the temporary area, and the storage data received from the outside and to be written to the disk device is redundant data. Without writing the temporary data to the temporary area, When the writing to the polar area is completed, a temporary area writing means for reporting the completion of writing to the outside, and a generating means for generating redundant data based on the stored data on the temporary area after the completion of the writing is reported. And a data area writing means for storing the redundant data in the redundant data area and writing the stored data in the stored data area.
[0012]
Further, when storage data to be stored is received from the outside, a plurality of the disk devices are selected, a temporary area in which the storage data is written is set as a temporary area in the disk device, and the same as each of the plurality of temporary regions. Means for writing the storage data in a multiplex manner, and means for reading the storage data from the temporary area when the storage data cannot be read from one of the disk apparatuses when reading the storage data from the temporary area. It was done.
[0013]
Further, when storage data to be stored is received from the outside, a temporary area in which the storage data is to be written is temporarily moved from a temporary area in a disk device different from the disk device having the storage data region in which the storage data is to be written. A means for selecting is provided.
[0014]
[Action]
The storage data to be stored from the outside is temporarily written to a temporary area without creating redundant data, and when this writing is completed, the completion of the writing is reported to the outside (for example, a host computer), and the storage data on the temporary area is stored. Is read out thereafter to generate redundant data, and the stored data and the redundant data are written to the respective data areas. This speeds up the response to the outside without using the cache memory.
[0015]
In addition, when a double write is performed in a temporary area on a plurality of disk devices, and thereafter a read request for the same storage data is received, if the storage data is read from the temporary area, the integrity of the storage data is improved. .
[0016]
At this time, in reading from the temporary area, a disk device that is not currently performing a read / write operation (this disk device is not called a drive busy) is selected from disk devices that are being double-written. By doing so, a delay due to drive busy can be avoided, and a high-speed write operation can be performed.
[0017]
When writing storage data from the temporary area to the storage data area, it is necessary to read out old storage data, new storage data and redundant data in the temporary area, and newly generate redundant data. When the temporary area in which the storage data is to be written is selected from a temporary area in a disk device different from the disk device having the storage data area to which the storage data is to be finally written, each drive can be multiplexed and stored in the temporary area. Speeds up writing to the data area.
[0018]
At this time, by selecting a temporary area to be selected from a disk device that is not currently performing a read / write operation, a delay due to drive busy can be avoided, and a high-speed write operation can be performed.
[0019]
【Example】
Hereinafter, an embodiment of a disk array device according to the present invention will be described.
[0020]
FIG. 1 is a diagram illustrating an overall configuration of a disk array device according to the present embodiment. In the figure, a disk array device 101 includes a control circuit 102, an upper interface circuit 103, a small disk interface circuit 104, and a magnetic disk device 111. The control circuit 102 includes a microprocessor 105, a program memory 106, a data buffer 107, a redundant information generation circuit 108, and a nonvolatile memory 109. The non-volatile memory 109 has a temporary area management table 110.
[0021]
FIG. 2 is a diagram showing a configuration of a magnetic disk device in the disk array device. All the magnetic disk devices 111 in the disk array device 101 are composed of a data area 201 for finally storing user data and a temporary area 202 for temporarily writing user data. The sizes of the data area 201 and the temporary area 202 are arbitrary.
[0022]
FIG. 3 is a software configuration diagram according to the present embodiment. This is executed by the microprocessor 105 shown in FIG. 1 and the program memory 106 in which the microprogram executed by the microprocessor 105 is stored. However, the generation of redundant data in the background data writing processing 304 is performed by the redundant information generation circuit 108. The scheduler 301 controls the entire disk array device 101. The scheduler 301 controls the host I / F control unit 302, the host command execution unit 303, and the background data write processing unit 304. The host command execution unit 303 executes the host command by using the read command processing 306, the write command processing 307, and the control command processing 308 through the command determination unit 305.
[0023]
FIG. 4 is a diagram illustrating a configuration of the temporary area management table 110 according to the present embodiment. In the figure, the management table includes a host device designation information 401, a control unit management information 402, a temporary area management information 403, and a next pointer 404 for confirming which of the temporary areas are used and which are free. It is configured. The next pointer 404 has two types of chains. One is a chain connecting used ones of the temporary area, and the other is a chain connecting vacant ones.
[0024]
The higher-level device designation information 401 includes a logical LBA (Logical Block Address) 405 and a logical LEN (LEN: Length) 406. The control unit management information 402 includes a physical drive No. 407, a physical LBA 408, and a physical LEN 409. The temporary area management information 403 includes a physical drive No. used as a temporary area. 410, a physical LBA 411 and a physical LEN 412 in the magnetic disk device. By checking the next pointer 404, it is possible to check the vacancy of the temporary area.
[0025]
Next, each process of receiving a write command from the host device and performing double writing on the temporary area will be described with reference to FIGS.
[0026]
Upon receiving the write data 502 from the host device 501, the control unit 503 first checks the block length of the write data 502 (701 in FIG. 7), and if the block length is larger than the threshold value, stores the block length in the temporary area. Instead, the normal write processing (702) is performed. That is, the write data 502 from the host device 501 is stored in the normal write area 508, and the created parity data is written in the normal parity write area 509. When the size of the write data 502 is large, the write penalty is small, so that it is stored directly in the final storage location.
[0027]
If the block length is smaller than the threshold value, a temporary area is secured using the next pointer 404 (FIG. 4) (703). When the data is secured, the data is managed so as not to overlap with a physical magnetic disk device in which data is actually written. By using different disk devices, when processing in the background, the new data in the temporary area and the old data at the position where the data is actually written are different disk devices, so they can be read in parallel. Accordingly, the processing for generating redundant data is sped up. Further, when securing a temporary area, a temporary area in which storage data is to be written is selected from a temporary area in a disk device that is not currently performing a read or write operation. By doing so, a delay due to drive busy can be avoided, and a high-speed write operation can be performed.
[0028]
If there is no free space in the temporary area (704), normal write processing (702) is performed. If an empty temporary area can be secured, double writing is performed on the areas 506/507 of the temporary area 511.
[0029]
Next, a process of reading user data that is double-written in the temporary area, generating parity data, and writing the parity data in the data area will be described with reference to FIGS.
[0030]
In the background processing, the control unit 603 first secures the address of the area 605/606 in which the already duplicated user data is stored from the temporary area management table 611 (801 in FIG. 8). Then, data is read from the temporary area on which the write processing has not been performed (802). In other words, if the magnetic disk device in which the area 606 storing data is stored in the area 605/606 in which the double-written user data is stored is used in the read processing, for example, Is read from the area 605 in which the data is stored. By doing so, a delay due to drive busy can be avoided, and a high-speed read operation can be performed.
[0031]
If data cannot be read from one temporary area due to a failure of the magnetic disk device (803), data is read from another double-written temporary area (804). As described above, due to the double writing, data can be retained even if the disk device fails, and data integrity is improved.
[0032]
Based on the data read from the temporary area and the parity data 608 that needs to be changed, new parity data is generated by the redundant information generation circuit 604 (805), and the data is written to 607 and the parity data is written to 608. (806). When writing storage data to the data area, if a write error occurs in a disk device having the storage data area, the disk device is automatically disconnected, and writing to the disk device is not performed thereafter.
[0033]
In the above embodiment, an example of a disk array device using a magnetic disk device as a storage medium has been described. However, the present invention can be similarly realized when another recording medium, for example, an optical disk is used.
[0034]
【The invention's effect】
According to the present invention, it is possible to prevent a decrease in write processing performance due to a so-called write penalty in RAID4 and RAID5, and to determine whether or not the system has a small number of writes to the disk device (a large amount of read processing). Regardless of this, it is possible to improve the performance of the system by using RAID.
[0035]
Further, the same interface as that of a conventional OS (Unix or the like) can be used, and there is an effect that an existing application can be used because an application is not made aware of a temporary area.
[Brief description of the drawings]
FIG. 1 is a block diagram of a disk array device according to an embodiment of the present invention.
FIG. 2 is an explanatory diagram of a storage area in a magnetic disk device according to one embodiment of the present invention.
FIG. 3 is a software configuration diagram according to an embodiment of the present invention.
FIG. 4 is an explanatory diagram of a temporary area management table according to an embodiment of the present invention.
FIG. 5 is an explanatory diagram of a write operation according to one embodiment of the present invention.
FIG. 6 is an explanatory diagram of a write operation in the background according to one embodiment of the present invention.
FIG. 7 is a flowchart of a process of writing data to a temporary area according to an embodiment of the present invention.
FIG. 8 is a flowchart of a process of writing data from a temporary area to a data area according to an embodiment of the present invention.
[Explanation of symbols]
101 Disk Array Device 102 Control Circuit 103 Upper Interface Circuit 104 Small Disk Interface Circuit 105 Microprocessor 106 Program Memory 107 Data Buffer 108 Redundancy Information Generation Circuit 109 Nonvolatile Memory 110 Temporary Area Management Table 111 Magnetic Disk Drive 201 Data Area 202 Temporary Area 301 scheduler 302 host I / F control unit 303 host device command execution unit 304 background data write processing 305 command determination unit 306 read command processing 307 write command processing 308 control command processing

Claims (4)

A plurality of disk devices, and at least one of the plurality of disk devices has a redundant data for reproducing the data that cannot be read when storage data written to another disk device cannot be read; A method of controlling a disk array device configured to be a disk device having a redundant data area for writing
In each of the at least two disk devices, an area for writing storage data to be stored is defined as a temporary area for temporarily writing the storage data, and a temporary area for finally writing the storage data written in the temporary area. Dividing into storage data areas ;
Before the storage data to be stored in the disk device received from the outside and the redundant data are created, the same storage data is once written in a multiplex manner in each of the plurality of temporary areas, and the writing to the temporary area is not performed. Upon completion, a temporary area writing step for reporting completion of writing to the outside,
A generating step of generating redundant data based on the stored data in the temporary area after the completion of the writing is reported;
A data area writing step of storing the redundant data in the redundant data area and writing the stored data to a storage data area in a disk device different from the disk device having the temporary area written in the temporary area writing step; A method for controlling a disk array device, comprising: 2. The method for controlling a disk array device according to claim 1, further comprising:
A method for controlling a disk array device, comprising: determining whether storage data to be stored in a disk device, which is received from outside, is longer than a predetermined data length. 2. The method for controlling a disk array device according to claim 1, wherein
A step of selecting a temporary area for writing the storage data from a temporary area on a disk device different from the disk device having the storage data area for writing the storage data when receiving storage data to be stored from outside; A method for controlling a disk array device having: 2. The method for controlling a disk array device according to claim 1, wherein
A method for controlling a disk array device, comprising the step of, when receiving storage data to be stored from the outside, selecting a temporary area in which the storage data is to be written, from a temporary area in a disk device that is not performing a read or write operation.

Images