JPH08263335A - Data storage - Google Patents

Abstract

PURPOSE: To arrange the files with small load an in response to the file access frequency despite the change of this frequency and to prevent the increase of seek value among the files by rearranging only the files that have high access frequency. CONSTITUTION: The access frequency is decided in every file unit consisting of plural blocks and stored, and the files which are correlative with each other and have their access frequency to be stored and higher than a prescribed level are extracted as one group. Then the block data on the extracted files are read out of the optical disks 11 and then successively written into the continuous or nearby blocks of these disks.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data storage device in which data is written to a data storage medium in block units and data written in blocks of the data storage medium is read out.

[0002]

2. Description of the Related Art In recent years, the amount of data to be managed in an office is increasing, and in order to manage such a large amount of data, a large-capacity rewritable storage medium (rewritable magneto-optical disk or magnetic In many cases, a collective optical disc device (auto changer) including a storage cell unit that stores a plurality of tapes, a plurality of optical disc drive units, and an accessor unit that moves an optical disc between the optical disc drive unit and the storage cell unit is used. Proposed.

However, the movement of the optical disk and the loading / unloading processing to / from the optical disk drive by the accessor section are extremely time-consuming processing as compared with other processing, and therefore, the "collective optical disk device control method" (Japanese Patent Laid-Open No. Hei 5).
No. 28615), there is proposed a device in which the storage position of the optical disk is devised so that the moving distance is minimized.

In this case, however, each optical disc handles data in block units, and at the time of writing, a new empty block is searched for and written, and the written data is read out in block units. ing. Therefore, while repeatedly writing and erasing data for each block, a plurality of blocks that make up one file are randomly fragmented (distributed), and even when the same file is continuously accessed. A seek operation (movement of the optical head) is required on the disk surface on the way, which causes a problem that the access speed is reduced.
In order to avoid this problem, it is necessary to relocate the blocks included in the same file so that they are contiguous and eliminate the fragment.

In order to eliminate fragments, processing is usually performed such that blocks included in the same file are found from information recorded in the file system and rearranged continuously. In the conventional system, blocks included in the same file are rearranged so as to preserve continuity as a file.

However, considering the position in the storage medium as the file after the rearrangement, the file arranged by paying attention only to the continuity has the continuity in the file, but as the file, the recording surface of the disc. Since it is recorded in a distributed manner on the top, when there are multiple files that are frequently accessed, the access positions on the disk are distributed over the entire disk surface, and seek operations between files are required for efficient access. There is a problem that you can not do.

[0007] In order to reduce the seek operation or seek amount between files, "USP53333311" predicts the possibility of being accessed for each file, based on which the priority is set for each file and blocks are relocated. In this case, the seek amount is reduced by performing a process of combining files having similar priorities.

However, the access frequency of a file is dynamically determined such that a block that has been frequently accessed until now is suddenly stopped or a block that has not been accessed until now is suddenly and frequently accessed. It changes, and it is uncertain whether or not files having the same access frequency relocated at a certain access frequency will be accessed to the same extent thereafter. When arranging a file considering only the access frequency for a certain time, the file allocation may change from the one considering the desired access frequency due to changes in the access frequency. In order to maintain the effect of, it becomes necessary to relocate all files according to the frequency change. Relocation of all files is a very heavy processing, and frequent execution of this processing leads to deterioration of the performance of the entire system.

Focusing on the frequency of seek operations between files, it is considered that seek operations between files that are frequently accessed occur more frequently. This shows that the seek amount can be reduced even if data relocation is performed only for files that are frequently accessed.

Further, even when considering the access frequency of the files in this way, the relationship between the files is ignored, so that files having similar access frequencies are also arranged disjointly within the range of files having the same access frequency. The seek amount becomes smaller, but seeks occur within that range. When the file access frequency changes and a frequently accessed file group is not necessarily recorded in a close place, the seek amount between files further increases.

When considering a collective storage device using a portable storage medium library device, the influence of the seek operation on the access speed becomes particularly remarkable. In the collective storage device using the library device, the location of the data is the location A. Location on storage medium loaded in drive device for reproducing data B. It is either on the storage medium stored in the storage medium storage unit of the library device. If the data is at location B, first remove the storage medium loaded in the drive device, move the storage medium containing the data to be accessed from the storage of the library device and load it into the drive, then read / write the data. Will be done. The time required for these processes is hundreds of times longer than that at the place A. The main causes are spin-down processing before removing the storage medium from the storage medium drive device, spin-up processing after loading, and the like. From this, it can be said that an important condition for high-speed access is that the data to be accessed is on the storage medium loaded in the storage medium drive device. However, in reality, even a block included in the same file is distributed over a plurality of optical disks by the operation of finding a new block during writing.

[0012]

As described above, according to the conventional method, files having similar access frequencies are randomly arranged. Therefore, the files are rearranged in the collective storage device equipped with the library device. In this case, if there are a plurality of storage media in which files having similar access frequencies are recorded, the files are distributed and recorded in the plurality of media. For example, if the files A and B are always accessed at the same time, and if the files are similarly distributed over a plurality of media, an access request to the non-loaded media is issued every time the worst accessed file changes. As a result, the medium is exchanged every time the file is accessed, and the access speed is extremely reduced.

In order to prevent such a situation, it is only necessary that the files accessed at the same time, regardless of the access frequency, exist on the same medium, or exist at close positions on the same storage medium. Conceivable. For that purpose, it is considered that not only the access frequency of the file but also the correlation information that the file B is highly likely to be accessed at the same time when a certain file A is accessed may be used.

According to the present invention, relocation is performed only for frequently accessed files, so that even if the access frequency changes, file placement can be performed according to the frequency with a small load, and the seek amount between files can be increased. The purpose is to prevent.

Further, in addition to the continuity / access frequency of blocks as information for block rearrangement, the interrelationship between blocks is used in combination, whereby more effective block rearrangement is performed and the access frequency is improved. The purpose is to prevent a decrease in access speed even when there is a change.

[0016]

In the data storage device of the present invention, data is written in block units to a data storage medium or data written in blocks of this data storage medium is read out. Determination means for determining the access frequency in file units composed of the plurality of blocks, storage means for storing the access frequency in file units determined by the determination means, and the access frequency stored in the storage means is a predetermined value. By the extracting means for extracting the plurality of files having the above-mentioned correlations as one group, the reading means for reading the data of each block of each file extracted by the extracting means from the data storage medium, and the reading means. The data of each block corresponding to each read file is stored continuously on the data storage medium. And a sequentially writing write unit block of the block or in the vicinity.

The aggregate type data processing device of the present invention comprises a plurality of data storage media in which data is written in block units and data written in this block is read out. Determination means for determining the access frequency of each file,
Storage means for storing the access frequency of each file determined by the determination means, extraction means for extracting a plurality of correlated files having a predetermined access frequency or more stored in the storage means as one group, Reading means for reading the data of each block of each file extracted by the extracting means from the plurality of data storage media;
And writing means for sequentially writing the data of each block corresponding to each file read by the reading means to a continuous block or a neighboring block on the one data storage medium.

In the data storage device of the present invention, data is written in block units to the data storage medium,
When data written in a block of the data storage medium is read out, a judging means for judging an access frequency in file units composed of the plurality of blocks, and an access frequency in file units judged by the judging means. Storing means for storing a plurality of files having an access frequency of a predetermined value or more stored in the storing means as a group, and data of each block of each file extracted by the extracting means It comprises a reading means for reading from the storage medium, and a writing means for sequentially writing the data of each block corresponding to each file read by the reading means to a continuous block or a neighboring block on the data storage medium. There is.

[0019]

According to the present invention, data is written in block units to a data storage medium, or data written in blocks of this data storage medium is read out. The access frequency of each unit is determined, the determined access frequency of each file is stored, and a plurality of correlated files having the stored access frequency of a predetermined value or more
Data of each block of each extracted file is read from the data storage medium, and the data of each block corresponding to each read file is a continuous block or a neighborhood of the data storage medium. The blocks are written sequentially.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a system configuration diagram of a collective optical disc device as a collective data processing device of the present invention.

This collective optical disk device is composed of a CPU 1, a main memory 2, a hard disk (HDD) 3, an autochanger control unit 4, an optical disk drive control unit 5, an autochanger 6, and a communication control unit 7.

The hard disk 3 and the main memory 2 are for storing programs and data. Further, a management table 20 described later is also stored.

The CPU 1 controls the entire operation, reads a program stored in the hard disk 3 into the main memory 2, and controls each unit according to the contents.

The autochanger 6 comprises a plurality of storage cells 8a to 8j, a plurality of optical disk drives 9a to 9c, and an accessor 10, and is controlled by the autochanger controller 4.

The storage cells 8a (8b to 8j) are for storing the optical disk 11 as a portable storage medium. The optical disk drives 9a, 9b, 9c are for writing and reading data to and from the optical disk 11. The optical disk drive control unit 5 controls the three optical disk drives 9a, 9b, 9c.

The accessor 10 includes an optical disk 11 between the storage cells 8a, ... And the optical disk drives 9a, 9b, 9c.
Is to move. The communication control unit 7 is connected to a LAN, receives a command sent from an external device through the LAN, and sends a processing result.

The respective units are connected by the system bus 12. The optical disk 11 is a magneto-optical disk, and it is possible not only to write-once write data but also to delete or change data.

The data in this device is managed in units of all blocks (1 Kbyte (K is 1024)). 1
The number of blocks of one optical disk 11 is 256K and 256
It has a capacity of M bytes, and as shown in FIG. 2, block numbers 1 to n are assigned to each block unit. A total of 10 optical disks 11 are managed, and the entire apparatus can manage 2.5 Gbytes of data, which is the capacity for 10 optical disks 11.

In the device, logical block numbers (1 to 2.5 M) are attached to the 2.5 M blocks (capacity of 2.5 GB) to be managed, and each logical block is A logical block / physical block management table 21, which will be described later, in the management table 20 manages the physical block number including the number of the optical disk 11 in which data is actually stored and the block number.

FIG. 3 shows a schematic configuration of software of the above-mentioned aggregate optical disk device. That is, the aggregate optical disc device includes a transfer control unit 31, an access information management unit 32, a rearrangement processing unit 33, a grouping processing unit 34,
Optical disk selection unit 35, group selection unit 36, library device control unit 37, autochanger 38, storage unit 3
9, a drive device 40, a transport unit 41, and a storage medium (medium) 42.

The transfer control unit 31 corresponds to the communication control unit 7, and the access information management unit 32, the relocation processing unit 33, the grouping processing unit 34, the optical disk selection unit 35, the group selection unit 36 are the CPU 1, the main memory. 2, the hard disk 3, the library device controller 37 corresponds to the autochanger controller 4 and the optical disk drive controller 5, the autochanger 38 corresponds to the autochanger 6, and the storage unit 39 corresponds to the storage cells 8a to 8j. Correspondingly, the drive device 40 corresponds to the optical disk drives 9a-9
Corresponding to c, the transport unit 41 corresponds to the accessor 10,
The storage medium 42 corresponds to the optical disc 11.

The autochanger 38 includes a storage medium 42.
A storage unit 39 for detachably storing a plurality of storage media, and a storage medium 42
Drive device 4 for loading and reading information
0, a storage medium 42 between the storage unit 39 and the drive device 40.
The optical disc device is a collective optical disc device having a transport unit 41 for transporting the optical disc. The storage medium 42 used in this device is portable and can be written as well as deleted.

As shown in FIG. 2, the storage area in the storage medium 42 is divided into blocks of a certain unit and managed.
A unique number is assigned to each storage medium 42 within the device.

All access to blocks in the collective optical disk device is performed via the transfer control unit 31. A storage medium 42 whose access is already loaded in the drive device 40
When it is for the upper block, the block is read from or written to the loaded storage medium 42. When an access to a block that is not loaded in the drive device 40 occurs, the library device control unit 37 loads a desired storage medium 42 into the drive device 40 and then accesses. When loading the storage medium 42 in the storage unit 39, it is necessary to move the already loaded storage medium 42 to the storage device, but which loaded storage medium 42 is to be stored is determined by the storage medium selection unit 35.

The access information management unit 32 records the access to the data in the collective optical disk device performed through the transfer control unit 31. Relocation of the data is requested by the host system or the access information management unit 3
2 is started when it is determined that the access frequency of a specific block has increased from the recorded access information. The data rearrangement processing is performed by the rearrangement processing unit 33, and the relevant data is rearranged at an appropriate position in the collective optical disc device.

It is the relocation processing unit 33 that determines an appropriate position for relocating the data. The relocation processing unit 33: 1) Since the blocks included in the same file are highly likely to be continuously accessed, they are continuously accessed on the same storage medium 42 in order to avoid a seek operation between blocks (movement of the optical head). It is desirable that it is recorded.

2) It is highly likely that a file with a high access frequency will be accessed with a high frequency in the future as well, so that the area on the loaded storage medium 42 or in the adjacent area on the storage medium 42 to reduce the seek amount. It is desirable to place the block distance value as a parameter.

3) Since a plurality of files having correlations with each other are likely to be accessed at the same time, the storage medium 4
In the same storage medium 42 in order to avoid the replacement operation of 2,
Further, it is desirable to arrange them in an adjacent area on the storage medium 42 in order to reduce the seek amount.

Arrangement of data is determined using a judgment criterion such as the above. Specifically, the grouping processing unit 34 uses the continuous information, access information, and file correlation information as files for data to be rearranged, and 1) grouping blocks included in the same file 2) having the same access frequency 3) Grouping files 3) Corresponding files are grouped, and blocks that are preferably arranged in adjacent areas on the same storage medium 42 are treated as a group so as to satisfy the above criteria as much as possible. Arrange.

When rearranging all the blocks in the aggregated optical disk device, the grouping selection unit 34 predicts the access frequency of the group with respect to the group obtained as a result of the grouping, and the group which is predicted to have a high access frequency. They are arranged in order from the storage medium 42.

When the target of relocation is limited to a group that is frequently accessed, grouping is performed only on blocks related to the block that is frequently accessed at that time, and the blocks that are grouped are different. The rearrangement is performed only when distributedly recorded on the storage medium 42 or at separate locations on the same storage medium 42. The groups that are frequently accessed are arranged in as continuous areas as possible on the same storage medium 42. When relocating a group continuously, it may be necessary to move the block where data is already recorded to secure the free area, but the access frequency and the file that is actively accessed change dynamically. Therefore, it is desirable to move the blocks in the group so as not to be distributed when the blocks to be moved are already grouped in case the blocks to be moved become actively accessed. For this reason, data movement for securing a free area is also performed in group units.

As a group to be moved to secure a free area, the grouping selection unit 34 selects a group in the storage medium 42 of the movement destination of the high-frequency access group to be relocated, which has a low access frequency. .

In the rearrangement, if there is not enough free area on the loaded storage medium 42, the storage medium selecting unit 35 selects the loaded storage medium 42 with low access frequency, and the storage medium having the empty area in the storage unit 39. The exchange processing with 42 is performed.

When the group to be moved is large and the relocation processing of the group has a very high load, the blocks to be grouped are limited to suppress the size of the group and then the relocation is performed.

[Embodiment 1] Next, the processing contents of the collective optical disk device will be described in detail. 4 to 7 show the management table 20 and data used in this embodiment.

The management table 20 comprises a logical / physical block management table 21, a file management table 22, an optical disk management table 23, a system control data storage table 24, and an optical disk block management table 25.

FIG. 4 shows a logical / physical corresponding block management table 21 for recording the correspondence of logical blocks / physical blocks of all 2.5M blocks in the collective optical disk device. An optical disk number and a physical block number of a corresponding physical block correspond to each logical block. 0 is recorded in the item of the physical block corresponding to the unrecorded logical block.

FIG. 5 shows an optical disk block management table 25 for managing blocks on the optical disk 11. The blocks on the optical disk 11 are managed by a bitmap. Each bit on the bitmap continuously corresponds to a block on the optical disc 11. A bit of 1 indicates that the corresponding block is used,
0 indicates that the corresponding block is not used.

FIG. 6 shows the contents of the file management table 22 for managing the files recorded in the collective optical disk device. The file number is used to identify the file. L and H (k) are areas used in the LRU-K algorithm described later to record the access frequency. The group number (column) is a number indicating the group to which the file belongs. A group number of 0 indicates that the file does not belong to any group. Group number 0 is also assigned to the newly created file. The block number indicates the logical block number of the recorded block of the file.

In this embodiment, file access information is recorded using the LRU-K algorithm. (EJ O
′ Neil and G. Weikum: The LRU-K Replacement Algo
rithmFor Database Disk Buffering: Proceedings of
1993 ACM SIGMOD pp. 297-306) LRU-K is an extension of the conventional LRU algorithm and uses the access history of data to judge the access frequency of data. K of LRU-K is an integer value, and indicates that the past K times of access time are stored for the data.

When recording the access history to data,
It is better to process access that occurs a plurality of times in a burst in a series of processing, such as retry of processing, reading to update data, and writing immediately after that, as one access. Therefore, the last access time (L) is recorded in each data in addition to the access times (H (n)) of the past n times. In the LRU-K algorithm, when data is accessed, first, it is confirmed that a certain time (correlation reference time) has elapsed from the last access time to the access. Accesses that occurred within the correlation reference time are not recorded in the history, only the last access time is updated. If the access is more than the correlation reference time, the current time is recorded as the previous access history H (1),
The previous access history is H (1) to H (2), H (2)
Is updated to H (3) by shifting backward one by one.

In this embodiment, the value of K is 2, and the last access time and the past two access histories are recorded for each data. 7 shows the location of the optical disk 11 (storage cell 8a, ... Or optical disk drive 9a, ...) In the collective optical disk device, 201 indicates the optical disk drive 9a, 101 indicates the storage cell 8a, 10
7 indicates a storage cell 8g) and an optical disk management table 23 for recording the number of empty blocks in the optical disk 11.
Is shown.

FIG. 8 is a system control data storage table 2 for storing control data used in other systems.
4 is shown. The correlation reference time value is a numerical value used in the LRU-K algorithm used for access information management. The access frequency judgment value is the exchanged optical disk selection unit 3
It is a numerical value used in 5 to determine whether or not data is frequently accessed. The block distance judgment value is a value used to judge whether blocks are dispersed in the same optical disc 11, and a value of 1 or more is recorded. It is assumed that blocks are dispersed when the difference between the maximum number and the minimum number of physical blocks included in a group when a certain group is given is larger than the number of blocks in the group times this value.

The processing of this embodiment will be described below with reference to FIGS.
This will be described with reference to the flowchart shown in. FIG. 9 is a diagram showing a flowchart for explaining the overall processing.

When the system is started, the CPU 1 first
The program and management information stored in the hard disk 3 are read into the main memory 12 and execution of the program is started (ST1).

A data access request and a control command are received from the communication control unit 7 via the LAN (ST2). When a request to access data is received, first LRU-K
Data access information is recorded using an algorithm (ST3), and request processing is started.

Whether the request is a data read request (ST4),
If the request is for overwriting existing data (ST5), if the data is present on the optical disk 11 loaded in the optical disk drives 9a, 9b, 9c (ST6), the request processing is performed (ST7).

The requested data is stored in the storage slots 8a, ...
If it is on the optical disk 11 stored in (S
T6), first, an optical disk exchange process (see FIG. 11) is performed (ST8), a desired optical disc 11 is loaded, and then a request process is performed (ST7).

If the request is a new write request (S
T9), an item for managing a new file is provided in the logical / physical block management table 21, an empty area on the loaded optical disk 11 is searched from the optical disk block management table 25 and the optical disk management table 23, and if there is a sufficient empty area, Writes data there (ST1
0).

If there is not enough free space on the loaded optical disk, the storage cells 8a to 8j of the autochanger 6 will be described.
Select an optical disk 11 having a sufficient free space from the optical disks 11 stored in
Writing is carried out after loading into b and 9c. Optical disc 1
For the exchange of No. 1, the optical disc exchange process (see FIG. 11) is performed.

When the request is a file deletion request (ST11), the logical / physical block management table 21
The file is deleted from the file, the block used by the file is searched from the file management table 22, and the optical disk block management table 25 for the searched block is searched.
The block is released by setting the bit in the block to 0 (ST1
2).

When the request processing is completed, the access frequency of the accessed file is checked (ST13), and if it is determined that the file is frequently accessed, the file grouping processing (ST14, see FIG. 12). I do.

In this embodiment, the data of LRU-K was used to determine whether the file was accessed frequently, and the difference between the time recorded in H (2) of the file and the current time was prepared. If the value is smaller than the access frequency judgment value (stored in the system control data storage table 24), it is judged that the file is frequently accessed.

As a result of the grouping processing, when it is determined that the blocks that are determined to belong to the same group are dispersed (when the blocks are recorded in a distributed manner on a plurality of optical disks 11 or when they are separated on the same optical disk 11). If it is recorded at a different position) (ST15), block rearrangement processing (ST16, see FIG. 13) is performed.

The logical / physical block management table 21, which is a logical block / physical block correspondence table, is used to determine whether or not the groups are dispersed. When the blocks included in the group are included in the plurality of optical disks 11, it is determined that the groups are dispersed.

The difference between the maximum physical block number and the minimum physical block number of the blocks included in the same optical disk 11 in the group is the block distance judgment value (> 1) (data for system control storage) in the number of blocks in the entire group. (Stored in table 24) is larger than the product of the groups, it is determined that the groups are dispersed.

If the received request is an end request (ST17), the system management data is saved in the hard disk 3 and the process ends (ST18). Figure 10 shows LR
It is a figure which shows the process which records access using a UK algorithm.

In order to record the access information to the file, first, the last access time L to the file recorded in the file management table 22 is read (ST20).
Time of access (current time) and last access time L
If the difference between is smaller than the correlation reference time (ST21),
Since this access is judged to be a part of burst access, the access is not recorded and the final access time L
Is replaced with the current time (ST22).

When the difference between the last access time and the current time is larger than the correlation reference time (ST21), H (i) for the accessed file recorded in the file management table 22 is updated (ST23). , 24, 2
5). H (i) is the time of the i-th access in the past from the most recent access. H (1) is updated at the time of this access, H (2) is updated to the value of H (1) before updating, and H (j) is updated to the value of H (j-1) before updating. . The value of H (K), which is the oldest access time, is updated to the value of H (K-1) before the update, and the original H
The value of (K) is discarded. The current time is written in the final access time L.

FIG. 11 is a diagram showing a process of exchanging the optical disk 11. In order to load the optical disk 11, the optical disks 11 loaded in the optical disk drives 9a, ...
It is necessary to select and take out one of the groups, but in this embodiment, the average of access frequencies of all the groups in the optical disk 11 is used for selection. That is, the optical disk 11 having the smallest average access frequency of the included groups is selected and set as an exchange target.

The access frequency to the group is assumed to be the same as the highest access frequency of the files included in the group. However, the access frequency of the group including the file whose correlation reference time has not elapsed since the last access time is set to H (K) regardless of the value of H (K) so that the optical disk 11 including the file just accessed is not removed as much as possible. Suppose it is the maximum.

First, the logical / physical block management table 2
1 and the file management table 22 to select all the groups included in the loaded optical disc 11.
A file management table 22 is provided for each selected group.
Of the files included in the group with reference to H (K)
Of the maximum value of, that is, the time of H (K) of the latest time of accessing K times before the files in the group is allocated as H (K) of the group (ST31).

Next, for each loaded optical disk 11, the H (K) assigned in step 31 of all the groups included in the optical disk 11 are averaged to obtain the H (K) of the optical disk 11 ( ST32).

The smallest H (K) of the optical disk 11,
That is, the optical disk 11 having the oldest access time K times before is selected as the optical disk 11 having the least possibility of being accessed in the future (ST33).

FIG. 12 is a diagram showing a data grouping process. The data to be grouped is close to the access time, and it is considered that there is a high possibility that the data will be accessed collectively in the future access.

In this embodiment, the file access frequency is used to group data. That is, files that have been accessed K times within the access frequency judgment value from the current time are grouped as high-frequency access files. However, the files to be grouped are loaded into the optical disc drives 9a, ... At the time of the grouping process in order to prevent the replacement of the optical disc 11 for the relocation in the rearrangement process after the grouping process. The file is limited to the file on the optical disk 11.

First, a new group number G that does not exist in the file management table 22 is created (ST41). Next, the file management table 22 is searched (ST42),
The difference between the current time and H (K) is smaller than the access frequency judgment value (stored in the system control data storage table 24), that is, K times or more are accessed within the time specified by the access frequency judgment value. The received file is selected, and the group number of the file recorded in the file management table 22 is changed to the new group number G (ST4
3). At this time, the logical / physical block management table 21 and the optical disk management table 23 are also referred to at the same time, and files including blocks on the unloaded optical disk 11 are not targeted for grouping.

FIG. 13 is a diagram showing an example of distribution of groups in the apparatus. FIG. 14 shows a rearrangement process for concentrating groups recorded on a plurality of optical discs dispersedly on one optical disc 11.

It is assumed that the files included in the group H are distributed and recorded on three optical disks H, L, and O (H (H), H (L), and H (O) in FIG. 13). Since it is highly likely that the files included in the group will be collectively accessed in the future, it is preferable that the files be recorded on the same optical disk 11 in order to reduce the number of times the optical disk 11 is exchanged.

In the rearrangement process, first, a free area of the optical disk 11 including a part of the group H is searched from the optical disk management table 23, and the free area E ( If there is an optical disc x having x) (ST51), H is recorded collectively on the optical disc 11 (ST52).

When there is no optical disk 11 capable of collectively recording the group H (ST51), it is necessary to create an empty area for recording the group H on an appropriate optical disk 11. The size of the free space that must be created depends on the size of the group H and the optical disc x
The size H (x) of a part of the group H included in the group H and the current free area E (x) are excluded. Files are moved between the loaded optical discs 11 in order to create an empty area. At this time, however, the files to be moved are likely to be accessed in group units in the future, so the movement is performed in groove units.

In order to minimize the size of the empty area to be created, H (X) + E from the loaded optical disk 11
The optical disc 11 having the largest (x) is selected and set as the optical disc 11 as the destination of the group H (ST53).

Next, with reference to the file management table 22, the group having the lowest access frequency is selected from the other groups recorded on the selected optical disk 11 (ST54). The access frequency of the group is the same as the one with the highest access frequency among the files included in the group, similar to the one used in the optical disk exchange process (see FIG. 11). The value of H (K) of the file recorded in the file management table 22 is used for the comparison of the access frequencies. The group with the smallest H (K) value of the assigned groups (older) is selected and is designated as group L.

The size L of the selected group L is obtained from the file management table 22, and a part H of the group H included in the optical disk H is larger than the size H of the group H.
If the sum of (H), the empty area E (H) on the optical disc H and the size L of the group L is small (ST55),
From the groups included in the optical disc H, the group having the next lowest access frequency to the group L is selected and added to the previously selected group L (ST56).

This step 56 is repeated to increase the size of the group L until L + E (H) <H-H (H) is satisfied.

Next, the destination to which the group L selected in step 56 is moved is determined. First, the optical disk 11H
From the loaded optical discs 11 other than the optical disc 1 and the size H (x) of a part of the group H included in the optical disc 11
The sum of the empty areas E (x) on 1 is larger than that of the group L selected in step 56 (ST5
7) Let it be an optical disc L. When a suitable optical disk 11 is found, it is saved in a part of the work area of the group H included in the optical disk L and added to the empty area. Then, the group L is moved from the optical disc H to the empty area of the optical disc L (ST58).

If no optical disk 11 having a sufficient area for moving the group L is found (ST5
7) Reduce the size of group L or
This is dealt with by reducing the size of.

First, in order to minimize the size of the group L, among the files included in the group L, the file with the least access frequency, that is, the file with the smallest H (K) value (old) is selected, and the group L is selected. Remove from L (ST59).

At this time, if L + E (H) <H-H (H) is not satisfied due to the size change of the group L (ST60), the file with the least access frequency is removed from the group H (ST61). ).

The steps 57 to 61 are repeated until the optical disk x satisfying H (x) + E (x) <L is found, and the found optical disk 11 is set as the optical disk L, and the group L is moved to the optical disk L (ST58), the group H. Is moved to the optical disc H (ST52).

FIG. 15 is a diagram showing a process of moving the group H. A part of the group H recorded in the optical disk 11H is read onto the work area and the original area is set as an empty area (ST71).

Next, the groups H are recorded in order from the block having the smallest block number in the empty area of the optical disc H (ST72). By doing so, the blocks included in the files in the group H are continuously recorded in the physical blocks having smaller numbers as they are closer to the head, and the optical disc drive 9a (9b, 9b, 9b) is accessed during file access.
There is no need to change the seek direction in c).

[Embodiment 2] For the grouping of files, the correlation information of the files is used together with the access frequency of the files to relocate all the blocks in the collective optical disk device.

FIG. 16 shows a flow chart for explaining the overall processing of this embodiment. Since the process of relocating the entire collective optical disk device is very heavy, the relocation process (ST81) that was automatically performed by monitoring the file access frequency in the first embodiment is requested by the host system. Only when (ST82). Other processes are the same as the processes in the first embodiment (FIG. 9).

FIG. 17 is a flow chart for explaining the rearrangement process for the entire apparatus. First, the number N of the optical disk 11 to be relocated is initialized to the smallest optical disk number, and the relocation free area of the optical disk 11N is initialized to the entire area of the optical disk 11 (ST91).

Next, a grouping process is performed on the entire files recorded by the method shown in FIG. 18 (ST92). When the grouping process for the entire device is completed, the access frequencies of the groups are compared by the same method as that used in FIG. 11, and the number of the group with the highest access frequency is set to G (ST
93).

By using the work area, the data recorded on the optical disc N and the data contained in the group G are swapped to write the group G from the head of the optical disc N (ST97 to 99). The data written on the optical disc N before the rearrangement moves to the original area of the group G. At this time, the storage area other than the newly rearranged area in the optical disc N is stored as the rearrangement free area of the optical disc N.

When the rearrangement free area of the optical disc N is smaller than the group G (ST94), the rearrangement target is moved to the next optical disc 11 (ST95), and the group G is recorded from the beginning of the next optical disc 11. To do.

When the movement of the group G is completed (ST9
7), the group having the highest access frequency next to G is selected and newly set as group G (ST98). Further, N, which is the number of the optical disk 11 of the moving destination, is initialized to the minimum optical disk number (ST99).

Until all groups have been moved, or the destination of group G cannot be found (ST
96), steps 94-99 are repeated. FIG. 18 is a diagram showing a file grouping process (grouping process) when the access frequency and the file correlation information are used for file grouping. Whether or not the files have a correlation is determined by the method shown in FIGS.

First, the current time is substituted for T (ST10
1), the group numbers of all files recorded in the file management table 22 are initialized to 0 (ST10).
2). Next, the newly issued group number is substituted for G (ST103).

The value of H (2) is checked for all the files in the file management table 22, and the group number of the file whose difference between T and H (2) is less than the access frequency judgment value is set to G (ST105). The value of H (2) is used for determining the access frequency, and by this, files having similar access frequencies are grouped.

Next, among the files that have already been grouped into G in step 105, a file having a correlation is searched by the method shown in FIGS. 19 to 21, and the group number of the file determined to be included in G Be G (ST
106).

When grouping for G is completed (ST
106), and reset T to (T-access judgment value) (S
(T107), the processes of steps 103 to 107 are repeated until all files in the apparatus are included in any group (ST104).

FIG. 19 shows a grouping process when grouping files that are accessed at the same time as having a correlation. Given a particular file,
The file is set as a transfer target file, and H (1) and H (2) of the transfer target file are recorded in Ha and Hb, respectively (S
T111).

Next, H (1) and H (2) of all ungrouped files in the aggregate optical disk device are compared with Ha and Hb, respectively, and the difference between H (1) and Ha is the correlation reference time. When a file whose value is the difference and the difference between H (2) and Hb is the correlation reference time value is found, the file is determined to be correlated and the same group number as the file to be moved is recorded (ST112, 113).

FIG. 20 shows a process for judging the correlation of files using directory information. When a specific file is given, information on the directory in which the file exists is obtained from the host system of the collective optical disk device via the transfer control unit 31 (ST121).

From the obtained information, files included in the same directory as the file to be moved are grouped as having a correlation (ST122, 123). FIG. 21 shows
This shows a process in which a file to be correlated is recorded in advance and the correlation specified at the time of rearrangement of blocks is considered.

FIG. 22 shows a file correlation recording table for recording the correlation of a previously designated file. When a specific file is given, the file correlation data shown in FIG. 22 is referred to, and the specified files correlated with the file are grouped (ST13).
1, 132).

[Third Embodiment] The flowchart of the entire processing is the same as that of the first embodiment shown in FIG. 9, but in the third embodiment, the correlation of files is used to group the files to be relocated. That is different. Similar to the first embodiment, only the group including the frequently accessed file is rearranged.

FIG. 23 shows the grouping process performed in the third embodiment. Similar to the first embodiment, the grouping process is started when the access information management unit 32 determines that the frequency of access to a particular file is high.

First, a new group number is issued so that it does not overlap with the group number in the apparatus and is set to G (ST14).
1). The file whose access frequency is determined to be high is set as the file to be moved, and the group number G is recorded in the file management table 22 (ST142).

H (2) for all files in the device
And the file whose difference between the current time and H (2) is smaller than the access frequency judgment value is regarded as a file having a high access frequency at that time and is grouped into the same group number G (ST143, 144). Fig. 19 ~ for all the converted files
A correlated file is searched by the method shown in FIG. 21 and included in the group G (ST145).

[Fourth Embodiment] The flow chart of the entire processing is the same as that of any one of the first to third embodiments, except that the data is managed in block units instead of file units in the apparatus.

The storage area is managed as a block in the collective optical disk device, and data transmission / reception with the host system is performed in block units. The block management table 51 is shown in FIG.
The access frequency is managed by using the LRU-K algorithm for each block. Also in FIG.
It does not have the logical / physical block management table 21 and the file management table 22 shown in FIG. The block management table 51 has, for each logical block number, a physical block number, a last access time (L), an access history H (1),
H (2), file number, order in file (number of pages, etc.)
, And the group number to which the file belongs are stored.

When the block rearrangement process is performed, the continuity as a file, the correlation information not stored in the collective optical disc device, and the like are obtained from the host system. Figure 25
Shows an example of the file management table 52 managed outside the collective optical disk device. The file management table 52 stores a logical block number string for each file number.

In response to a request from the collective optical disk device, the host system refers to the file management table 52 and sends file information and directory information to the collective optical disk device. Since the block managed by the host system is a logical block, the aggregate optical disk device can transmit and receive data in response to a request from the host system while the physical block rearrangement process is being performed in block units.

[0118]

As described above in detail, according to the present invention,
By rearranging only the files that are likely to be accessed, it is possible to bring out the effect of reducing the seek between files by rearranging while suppressing the load of the rearrangement processing. In addition, the logical block / physical block conversion table is provided in the aggregate optical disk device, and the rearrangement process is performed in parallel with the data access service by eliminating the influence of the rearrangement process on the management table of the external system. Therefore, it is possible to effectively follow the change of the access frequency of the file which changes dynamically and to perform the effective block arrangement. This dynamic relocation method also has the effect of preventing file fragmentation due to access.

Further, by taking the correlation information of the files into consideration when relocating the files, the files having the correlation are collectively recorded, and even if the access changes infrequently, the cohesion of the files by the correlation is maintained so that the files can be accessed again. Even if the frequency becomes high, effective file arrangement can be realized without performing the rearrangement process again. This is particularly effective in a library device that uses a portable storage medium, because files that are considered to be accessed at the same time are stored in the same storage medium, so that storage medium exchange processing is suppressed.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of a collective optical disc device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a storage area of each block unit in the optical disc.

FIG. 3 is a block diagram showing the overall structure of a collective optical disc device.

FIG. 4 is a diagram showing a storage example of a logical / physical block management table for managing the correspondence between logical blocks and physical blocks.

FIG. 5 is a diagram showing an example of storage of an optical disk block management table for managing the usage status of physical blocks.

FIG. 6 is a diagram showing a storage example of a file management table for managing files.

FIG. 7 is a diagram showing a storage example of an optical disc management table for managing optical discs.

FIG. 8 is a diagram showing a storage example of a system control data storage table that holds data used for system control.

FIG. 9 is a flowchart for explaining processing of the entire apparatus.

FIG. 10 is a flowchart illustrating a process of recording an access frequency using the LRU-K algorithm.

FIG. 11 is a flowchart illustrating an optical disc exchange process.

FIG. 12 is a flowchart for explaining data grouping processing.

FIG. 13 is a diagram for explaining a group rearrangement process.

FIG. 14 is a flowchart illustrating a group rearrangement process.

FIG. 15 is a flowchart illustrating a group rearrangement process.

FIG. 16 is a flowchart for explaining overall processing according to the second embodiment.

FIG. 17 is a flowchart for explaining block rearrangement processing in the entire apparatus.

FIG. 18 is a flowchart for explaining grouping processing in the entire device using the correlation information of files.

FIG. 19 is a flowchart for explaining a grouping process of correlated files using reference time as correlation information.

FIG. 20 is a flowchart illustrating a grouping process of correlated files using directory information as correlation information.

FIG. 21 is a flowchart illustrating a process of grouping files by specifying a correlation between files.

FIG. 22 is a diagram showing a storage example of a table for recording the correlation of a specified file.

FIG. 23 is a flowchart for explaining a grouping process for a specific file using the correlation information of the file.

FIG. 24 is a diagram showing a storage example of a block management table of the present invention having a block interface.

FIG. 25 is a diagram showing a storage example of a file management table in a host system having a block interface.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... CPU 2 ... Main memory 3 ... Hard disk 4 ... Auto changer control part 5 ... Optical disk drive control part 6 ... Auto changer 7 ... Communication control part 8a-8j ... Storage cells 9a, 9b, 9c ... Optical disk drive 11 ... Optical disk 20 ... Management table

Claims (6)

[Claims] 1. A data storage device in which data is written in blocks to a data storage medium or data written in blocks of the data storage medium is read out, and a file composed of the plurality of blocks. Judgment means for judging the access frequency in units, storage means for storing the access frequency in file units judged by the judgment means, and a plurality of access frequencies stored in the storage means having a predetermined value or more and having a correlation Corresponding to the files read by the reading means, and the reading means for reading the data of each block of each file extracted by the extracting means from the data storage medium. The data of each block to Data storage device, characterized by comprising a writing means for sequentially written to click, the. 2. The plurality of correlated files comprises:
The data storage device according to claim 1, wherein the files are accessed at the same time. 3. The plurality of correlated files comprises:
The data storage device according to claim 1, wherein the data storage devices are files included in the same directory. 4. The plurality of correlated files comprises:
The data storage device according to claim 1, wherein the data storage device is a file group designated by an external device. 5. A collective data processing device comprising a plurality of data storage media in which data is written in block units and data written in the blocks is read out, in file units constituted by the plurality of blocks. Determination means for determining the access frequency of the storage unit, storage means for storing the access frequency for each file determined by the determination means, and a plurality of correlated access frequencies stored in the storage means that are greater than or equal to a predetermined value. Extracting means for extracting the files as one group, reading means for reading the data of each block of each file extracted by the extracting means from the plurality of data storage media, and each of the files read by the reading means The data of each corresponding block is transferred to successive blocks or near blocks on the one data storage medium. Data storage device being characterized in that anda writing means for sequentially written into the block. 6. A data storage device in which data is written in block units to a data storage medium or data written in blocks of the data storage medium is read out, and a file composed of the plurality of blocks. Judgment means for judging the access frequency in units, storage means for storing the access frequency in file units judged by this judgment means, and a plurality of files having an access frequency stored in the storage means of a predetermined value or more Extracting means for extracting as a group, reading means for reading the data of each block of each file extracted by this extracting means from the data storage medium, and data of each block corresponding to each file read by this reading means To sequentially write data to a continuous block or a nearby block on the data storage medium. Data storage device, characterized by comprising a means.