JPH07146757A - Disk device - Google Patents

Abstract

PURPOSE:To shorten average seek time by adjacently arranging areas where accesses are frequently generated with each other. CONSTITUTION:The definition information of a virtual disk is assigned in order of sub-disks Sd-a1 to Sd-a5 with a prescribed width from the leading area of a physical disk. As a result of the statistics and investigation of the number of times of access at the time of an operation as for each of these sub-disks, accesses are frequently generated in order of sub-disks Sd-a1, Sd-a5, Sd-a2, Sd-a3 and Sd-4, for instance. Because sub-disks having a high access frequency with each other are adjacently arranged and are arranged at the center of the disk, the redefinition information of the virtual disk is organized in order of Sd-a3, Sd-a5, Sd-a1, Sd-a2 and Sd-a4. Therefore, the distance on the physical disk of the Sd-a1 and Sd-a5 with large access frequencies is shortened and a probability that a head stands by at the central part of the disk becomes high.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk device, and more particularly to a method for shortening the average seek time when accessing a disk device using a virtual disk.

[0002]

2. Description of the Related Art Conventionally, various proposals have been made as a method of reducing an average seek time when accessing a disk device. For example, in a fixed disk device, the head position is stopped on the track that has been accessed immediately before, and the head movement amount is calculated for each access to calculate the head movement amount. A method in which the head is made to stand by in advance in the area (Japanese Patent Application Laid-Open No. 62-2
No. 23866), and a method of providing a plurality of heads to speed up access (Japanese Patent Laid-Open No. 2-218055).

[0003]

In the conventional seek time shortening method of this type, the heads are all controlled, and if there are many frequently accessed areas and the number of heads is larger than the number of heads, the physical control is performed. There is a problem in that the effect depends on the usage pattern of the area on the disc, which is not a preferable method.

It is an object of the present invention to provide a disk device in which the average seek time is shortened by arranging frequently accessed areas adjacent to each other using the virtual disk technology.

[0005]

DISCLOSURE OF THE INVENTION A disk device according to the present invention comprises a statistical processing means for performing statistical processing of each access state of a plurality of subdisks constituting a virtual disk, and at least 2 times of frequent access based on this statistical processing. Reorganizing means for reorganizing the virtual disks so that the two subdisks are arranged adjacent to each other.

Particularly, the sub-disks which are frequently accessed are characterized by being arranged near the center of the physical disk.

[0007]

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

FIG. 1 is a diagram showing the configuration of a virtual disk and an entity on a physical disk used in the embodiment of the present invention. As shown in the figure, the virtual disk vol-1 is a plex (p
lex) 1 has a physical disk that is an entity, and has five subdisks sd-a on this physical disk.
1, sd-a2, sd-a3, sd-a4, sd-a5
It is composed by.

Here, the physical disk is an actual disk device having sectors, cylinders, etc. Since the physical disk is usually treated as a file system, it is divided into arbitrary continuous areas for use. What corresponds to this continuous area is composed of an aggregate of existing areas (subdisks) of arbitrary size at arbitrary positions on the physical disk is called a plex, and is indicated by plex1 in FIG. .

This plex is associated with virtual disks of the same size so that the operator can access the virtual disks without being aware of the configuration on the physical disk. Further, the virtual disk can have a one-to-many relationship with the plex, and this association and its cancellation are in a dynamic relationship. It should be noted that this association is referred to as associate, and the cancellation thereof is referred to as associate.

Therefore, in the present invention, each subdisk sd-a of the physical disk is used by using the concept of the virtual disk.
It is assumed that 1 to sd-a5 are first defined (before the operation is started) as shown in FIG. At this time, the virtual disk definition information is allocated in the order of sd-a1 to sd-a5 with a width of 8192 blocks from the top area of the physical disk.

For each of these subdisks,
The number of accesses, which is input / output information during operation, is statistically surveyed, and as a result of this statistical survey, sd-a1, sd-a5
It is assumed that accesses are frequently made in the order of sd-a2, sd-a3, and sd-a4.

Sd-a1 and sd- that are frequently accessed
Since a5 is located at both ends (inner circumference and outer circumference) of the disk, it can be considered that it is a factor that increases the average seek time. Therefore, the contents of vol-1 are temporarily saved to another medium, and sd-b1 and sd as shown in FIG.
The plex 2 consisting of -b2, sd-b3, sd-b4, sd-b5 is newly defined and reorganized.

In FIG. 2, the sub disk number sd-b
1 to sd-b5 are sub-disk numbers sd-1 to sd-1 of FIG.
It corresponds to sd-5, respectively.

As shown in FIG. 2, the virtual disk redefinition information is such that subdisks with high access frequency are arranged adjacent to each other and at the center of the disk, and the access frequency for other subdisks. It is arranged from the center to the outside in order from the largest one.

By returning the contents of plex 1 to plex 2, which is the definition information of the virtual disk, and making it the entity of the virtual disk, the access to sd-a1 which was the access near the beginning of the disk is the center of the disk. Change to access to nearby sd-b1. Therefore, the distances of the frequently accessed sd-a1 and sd-a5 on the physical disk are shortened, and the probability that the head waits at the center of the disk is increased, so that the average seek time can be shortened. is there.

During disk operation, the number of times of access to each subdisk is always statisticized, and if the access frequency changes, new virtual disk definition information is generated and disk reorganization is performed to constantly seek. The time can be shortened.

FIG. 3 is a flowchart showing the process of reorganizing the virtual disk definition information during disk operation. In step 31, the subdisk,
The plex configuration is defined and is the first state shown in FIG.

When the operation is started in this state, in step 32, the access state (number of times) of each operating sub-disk is collected and statistical processing is performed. After the operation is performed to some extent and the access frequency for each sub-disk is calculated, it is determined in step 33 whether the position of the sub-disk to be accessed at that time is appropriate on the disk.

As the determination method, for example, the seek time of the head at the time of the access, the head seek movement distance, etc. are compared with a preset threshold value. Judge as inappropriate.

If it is not suitable, the process proceeds to step 34, and based on the statistical data of the access frequency of each subdisk,
The virtual disk definition information as shown in FIG. 2 is generated, the disk is reorganized, and the operation is restarted. Step 3
Return to 2. If it is determined to be appropriate in step 33, the operation is performed as it is and the process returns to step 32.

Although the operation is suspended for reorganization of the disk, in the case of a disk device having a duplicated structure, if the other disk is operated and used during the reorganization process of one disk, The operation interruption can be eliminated.

[0023]

As described above, according to the present invention, the frequently accessed areas on the physical disk are physically arranged adjacent to each other, and further, these areas are arranged at the center of the disk. There is an effect that the average seek time can be shortened regardless of the usage form of the physical disk area.

[Brief description of drawings]

FIG. 1 is a diagram showing a relationship diagram between a virtual disk and an entity on a physical disk and an example of virtual disk definition information according to an embodiment of the present invention.

FIG. 2 is a diagram showing a case where the arrangement on the disk is rearranged in the order of increasing access frequency.

FIG. 3 is a flowchart showing an outline of operation of the embodiment of the present invention.

[Explanation of symbols]

 vol-1 virtual disk plex1,2 physical disk sd-a1 to a5, sd-b1 to b5 sub disk

Claims (4)

[Claims] 1. A statistical processing means for performing statistical processing of respective access states of a plurality of subdisks constituting a virtual disk, and at least two subdisks having a large number of accesses based on the statistical processing are arranged so as to be adjacent to each other. A disk device comprising a reorganization unit for reorganizing a disk. 2. The disk device according to claim 1, wherein the reorganization unit is configured to reorganize the virtual disk when a head movement distance at the time of access exceeds a predetermined value. . 3. The disk device according to claim 1, wherein the reorganization unit is configured to arrange at least two sub-disks that have been frequently accessed near a disk center. 4. A disk device having a duplicated structure, wherein the reorganizing means is configured to currently use another virtual disk during the reorganization process.
The disk device according to any one of 3 to 3.