JP2002116936A - Disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the response performance of a disk array device by uniformizing loads based on read requests to plural HDDs (Hard Disk Drives). SOLUTION: A processor 10 counts the number of times of access required for the read request from a computer for each of the HDDs 12-112-3 and preserves it in an access frequency table 11. Then, the load condition of each of the HDDs 12-1-12-3 is confirmed from the access frequency table 11 and the copy of data stored in the most loaded HDD is prepared in a spare HDD 12-4.

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 which is connected to a computer via a bus and stores data used by the computer.

[0002]

2. Description of the Related Art A disk array device includes a plurality of hard disk drives (hereinafter, referred to as HDDs), and stores data relating to a write request from a computer into fixed-length striped data, and stores the data. In response to a read request, the striped data is restored to the original data and transmitted to the computer. For example, RA
ID (Redundant Arrays of Independent Disks)
In a method referred to as No. 3, a plurality of HDDs that store striped data in parallel and a parity data that is obtained based on the data stored in parallel are stored.
And one HDD. In a method called RAID5, all the constituent HDDs store striped data and parity data. As described above, in the disk array device, the striped data is distributed and stored in a plurality of HDDs.
Since it is possible to access another HDD in parallel with the access to, the response to the access from the computer is assumed to be high.

[0003]

However, the disk array device as described above still has problems to be solved. That is, the response performance of the disk array device described in the related art is merely a theoretical content, and access is concentrated on a specific HDD depending on the arrangement of striped data.
In some cases, desired response performance cannot be obtained. To solve such a problem, it is conceivable to arrange striped data so that accesses from computers are equalized. However, it is extremely difficult to predict that accesses from computers will be equalized. Therefore, an object of the present invention is to provide a disk array device in which the response performance from a computer is improved without considering the arrangement of striped data.

[0004]

In order to solve the above-mentioned problems, a disk array device according to the present invention comprises a plurality of hard disks for storing data used by a computer and at least one backup hard disk. In the disk array device, an access frequency table that holds the number of times read requests have been accepted for each of the plurality of hard disks, and a hard disk with the largest number of accesses is obtained by referring to the access frequency table, and a copy of data stored in the hard disk is copied. A processor for creating the backup hard disk. With such a configuration,
It becomes possible to equalize the access according to the read request from the computer in each disk. Further, in the disk array device according to the second aspect, a plurality of hard disks for storing data used by a computer,
In a disk array device having one backup hard disk, an access frequency table holding the number of read requests accepted for each of the plurality of hard disks, and the number of accesses per unit time based on the number of accesses held in the access frequency table. And a processor for making a copy of the data stored on the hard disk that is most frequently obtained on the backup hard disk.

[0005] With such a configuration, it is possible to equalize the access required for the read request from the computer in each disk. Since the load status is determined based on the access frequency per unit time, it is possible to perform the access leveling more accurately. Further, in the disk array device according to the third aspect, when the disk array device is provided with two or more backup hard disks, the processor sequentially selects a hard disk having a large number of accesses to create a copy. And With such a configuration, the access frequency can be further leveled.

Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a diagram showing a configuration of a disk array device 1 according to an embodiment of the present invention. The disk array device 1 receives an access request from a host computer (not shown) and a processor 10 for controlling the entire device. An access frequency table 11 for storing frequencies, four HDDs 12-1 to 12-3 for storing data accessed by the computer, and an HDD 12-
It has an HDD 12-4 used at the time of a failure of 1-12-3 and an interface 13 for receiving an access request from the computer.

In this embodiment, three HDDs 1
2-1 to 12-3 store the striped data in a distributed manner, and the HDD 12-4 has three HDs.
It is assumed that the same data as the data stored in one of D12-1 to 12-3 is stored. FIG. 2 shows the structure of the access frequency table, in which three HDDs are used.
12 shows a relationship between 12-1 to 12-3 and the number of times of read access to each HDD. Next, the operation of the disk array device having the above configuration will be described. First, when the disk array device 1 is started, an initialization process is executed to enable access from the host computer, and a write request and a read request are received via the interface 13. Then, the types of the received write request and read request are confirmed by the processor 10, and a write process or a read process for the HDDs 12-1 to 12-3, and an update process of the access frequency table 11 are performed as necessary. . At this time,
The processing of the processor 10 is shown in FIG. The processor 10 monitors whether a write request and a read request have been received from the computer (step 1), and upon reception, determines whether the request is a write request or a read request. (Step 2) If it is determined in step 2 that the received request is a write request, a write process is performed on the HDDs 12-1 to 12-3 to complete a series of processes.

[0007] On the other hand, if the received request is a read request, an HDD that stores data relating to the read request is selected from the HDDs 12-1 to 12-3, and a process of reading the corresponding data is executed. (Step 3),
The read data is transmitted to the computer via the interface 13. Then, when the processor 10 executes the reading process, the updating process of the access frequency table 11 is executed (step 4). The update processing of the access frequency table 11 is performed by counting up the number of accesses to the HDD storing the read data. For example, as shown in FIG. 4, the state of the access frequency table 11 before accepting a read request from a computer indicates that the number of read accesses to the HDD 12-1 is five, the HDD 12-2 is three, and the HDD 12-3 is three. It is assumed that there is a read request for the data stored in the HDD 12-1 and the HDD 12-2.
1 is updated to the value shown in FIG. That is, the HDD 12-
The number of accesses to HDD 1 from 5 to 6
The number of accesses to each HD is updated from three times to four times. As a result, the access frequency table 11 shows that each HD
The load conditions of D12-1 to D12-1 are held.

The processor 10 periodically checks the access frequency table 11 to see if there is any HDD which is deteriorating the response performance to the computer, and if necessary, a spare (hot spare disk) HDD 12-4.
Then, the data of the HDD under load is copied.
The determination as to whether or not a certain HDD is under load is made based on the number of accesses, and the HDD with the highest number of accesses is identified as having a high load. As a load determination criterion, the number of accesses per unit time is obtained, and a load having a larger value may be determined as a high load, or a determination may be made based on the amount of data read. . However, when determining the load based on the data amount, the data amount is held in the access frequency table 11. The load status of the HDDs 12-1 to 12-3 is confirmed by the processor 10 as described above, and the most loaded HDD will be described based on the above example.
The DD 12-1 determines that the load is the highest, and copies a copy of the data stored in the HDD 12-1 to the HD 12-1.
Create it in D12-4. In this way, by making a copy of the data stored in the HDD 12-1 which has the highest load on the spare HDD 12-4, the computer issues a read request to the HDD 12-1.
D12-1 and 12-4.
It is possible to equalize the access frequency based on a read request to each of the HDDs 12-1 to 12-4.

In the above description, it has been described that a high-load HDD is detected based on the number of accesses for a read request to the three HDDs, the number of accesses per unit time, and the like. Is H
Since it is used as a backup when the DDs 12-1 to 12-3 fail, it is desirable to reduce the frequency of use as much as possible. For this reason, instead of simply making a copy based on the number of accesses, a copy of the HDD with the largest number of accesses is copied from a copy exceeding a certain number of accesses (same even when the number of accesses per unit time is used as a reference). May be created. Further, in the above description, an example is described in which one spare HDD is provided. However, a plurality of spare HDDs may be provided and copies may be created for a plurality of HDDs under load. Further, the HDD to be copied may be periodically changed based on the state of the access frequency table 11.

[0010]

As described above, according to the present invention, a copy of an HDD under load is created in a spare HDD, and the access required for a read request is distributed to two HDDs. The response performance of the device can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a disk array device of the present invention.

FIG. 2 is a diagram showing a configuration of an access frequency table.

FIG. 3 is a flowchart showing an access frequency table update process.

FIG. 4 is a diagram showing a state of an access frequency table.

FIG. 5 is a diagram showing a state of an access frequency table.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Disk array apparatus 10 ... Processor 11 ... Access frequency table 12-1 to 12-4 ... Hard disk 13 ... Interface

Claims (3)

[Claims] 1. An access frequency table for holding the number of times of receiving read requests for each of a plurality of hard disks in a disk array device having a plurality of hard disks for storing data used in a computer and at least one backup hard disk. A disk array device comprising: a processor that refers to the access frequency table, finds a hard disk having the largest number of accesses, and creates a copy of data stored in the hard disk on the backup hard disk. 2. An access frequency table which holds the number of times each of a plurality of hard disks has received a read request in a disk array device having a plurality of hard disks for storing data used by a computer and at least one backup hard disk. And a processor that obtains the number of accesses per unit time from the number of accesses held in the access frequency table, and creates a copy of data stored on the most frequent hard disk on the backup hard disk. A disk array device characterized by the above-mentioned. 3. The disk array according to claim 1, wherein when two or more backup hard disks are provided, the processor sequentially selects a hard disk having a large number of accesses to create a copy. apparatus.