JPH0793105A - Information storage device - Google Patents

Abstract

(57) [Summary] [Purpose] To speed up data exchange processing by eliminating the need for parity regeneration associated with data exchange processing between recording media. [Structure] Data is exchanged between optical disks only between the same block addresses. That is, the data is exchanged within a range in which only the arrangement is changed without changing the combination of the block data in one RAID group.
Therefore, it is not necessary to rewrite the parity data P0. Further, the storage position of the file data is controlled so that the start addresses of the respective file data stored on the individual optical discs match the start addresses or the unused portions of the file data stored on the other three optical discs. As a result, when one file data is intensively written on one optical disc, it is possible to prevent the file data from being dispersed due to data replacement.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information storage device for storing information in a plurality of storage media by a RAID system, for example.

[0002]

2. Description of the Related Art In recent years, RAI has been used as a method for managing a large amount of data in an office at high speed and with high reliability.
The D (Redundant Arrays of Inexpensive Disks) method has been proposed. This technique is described in JP-A-2-236714, the interface April 1993, Nikkei Computer (1992.3.9) and the like.

FIG. 8 is a diagram showing the basic concept of RAID when writing data. In the figure, 1, 2,
Reference numerals 3, 4, and 5 are recording media such as optical disks forming a RAID, and each recording medium is set in a separate optical disk drive. Reference numeral 6 is a data generation unit that generates data to be recorded in RAID. Data generator 6
When the write data is received from the host, the write data distributor 6a divides the write data into four, and the five recording media 1 to
Data is written to the same block address on four recording media 1 to 4 out of 5. At this time, the parameter generation unit 6b in the data generation unit 6 divides the data D into four pieces.
Generate parity data P from 0, D1, D2, D3,
This is written in the same block address on the remaining one recording medium 5.

Here, the parity data P is obtained by P = D0 xor D1 xor D2 xor D3 (xor is an exclusive OR).

Therefore, even if an error occurs in the reading of data from a certain recording medium, it is possible to generate the data in which the reading error occurred from the data of the other four recording media of the same RAID group. For example, when the data D3 of the recording medium 4 is generated, it can be obtained by D3 = D0 xor D1 xor D2 xor P.

Although five recording media are used in this example, at least three recording media may be prepared, two of which are for data recording and one is for parity.

By the way, in an information storage device in which recording media can be freely exchanged with respect to a plurality of optical disc drives, when the recording medium recording the data to be accessed is not loaded in any optical disc drive, the recording medium from one of the optical disc drives is changed. It is necessary to remove and replace the target recording medium. At this time, if frequently-accessed data is recorded on the taken-out recording medium, this recording medium will have to be reloaded again in the near future. Therefore, a method is adopted in which data having a low access frequency is concentrated on one recording medium and is removed from the optical disk drive. As a result, the decrease in access speed of the entire system is suppressed.

Now, in order to concentrate the data with low access frequency on one recording medium, it is necessary to swap (replace) the data between the recording media. At that time, conventionally, for example, as shown in FIG. 9, the recording medium odn = 0 to 3
Even when one RAID group is formed by the above, the two data D02 and D13 are exchanged across different block addresses between the recording media. For this reason, each block No. in which the data has been replaced is It was necessary to rewrite the parity data P0 and P1. At this time, the rewritten parity data P0 'and P1' are obtained by P0 '= P0 + D02 + D13 P1' = P1 + D02 + D13.

[0009]

As described above, conventionally, it is necessary to regenerate and rewrite the parity data of each block in which the data is exchanged every time the file data is exchanged between the recording media. Therefore, it takes a long time to replace the file data.

The present invention is intended to solve such a problem, and eliminates the need for parity regeneration associated with the process of exchanging data between recording media, thereby making it possible to speed up the process of exchanging data. The purpose is to provide the device.

[0011]

In order to achieve the above-mentioned object, an information storage device of the present invention stores file data in a block unit in a plurality of first storage media and stores the file data in another second storage medium. In an information storage device that stores parity data for the same block of data in each first storage medium, replacement of file data in block units between the plurality of first storage media is performed in each of the first storage media. This is done between the same block addresses.

[0012]

In the information storage device of the present invention, a block forming one RAID is performed by exchanging file data in block units among a plurality of first storage media between the same block addresses of the respective first storage media. No changes are made to the combination of data. Therefore, it is not necessary to regenerate the parity data by this block data exchange, and the data exchange processing speed can be improved.

[0013]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the overall configuration of a RAID system according to an embodiment of the present invention.

As shown in FIG. 1, the RAID system is a host computer 10 that handles various file information.
1 and five optical disk drives (O
DD0) 102 to (ODD4) 106, a plurality (a large number) of optical disks 107 selectively loaded in the optical disk drives 102 to 106, and an auxiliary storage unit 108.
Optical disk drives 102 to 106, auxiliary storage unit 10
8. Data transfer between other parts management unit (CPU)
A DMA 109 that is independent of 111, a data generator 110 that distributes data written to the optical disc 107 by each of the optical disc drives 102 to 106, generates parity, and further generates read data when a read error occurs. Management unit 1 that comprehensively manages and controls each unit
And 11. The host computer 101, the optical disk drives 102 to 106, the auxiliary storage unit 108, the DMA 109, the data generation unit 110, and the management unit 111 are connected to each other via a bus 112.

A plurality (a large number) of optical discs 107 are accommodated in an accommodating portion (not shown), and movement of the optical disc 107 between the accommodating portion and each of the optical disc drives 102 to 106 is performed by an auto changer mechanism (not shown). That is, the optical disc 107 can be freely loaded into each of the optical disc drives 102 to 106, and the loaded optical disc 107 can be freely replaced.

This system comprises a host computer 10
It can be seen as a single optical disk drive system from 1 and has a structure in which file data can be read / written to / from a plurality of optical disks under RAID management in the same manner as read / write to a single drive.

FIG. 2 is a diagram relating to management of file data. In the figure, 201 is a file data management table. In the file data management table 201, the file No. (FN) and the NO. (Odn) and block NO. (Obn) is registered. For example, for a file with FN = 0, odn =
0 = 1 optical disk, each block of obn = 1 and 2, and od
optical disc of n = 1 obn = 15, 16, 17, 18
It can be seen that it is stored in each block of the number. The contents of the file data management table 201 are the management unit 11
Referenced and updated by 1. For example, the management unit 111
When receiving an access request from the host computer 101, the file data management table 201 is used to determine which block of which optical disk the file to be accessed is stored. In addition, the management unit 11
1 updates the contents of the file data management table 201 when data is exchanged between optical disks.

FIG. 3 is a diagram relating to the management of the access frequency of each file data. In the figure, 301, 30
2, 303 and 304 are four optical discs (odd) except the optical disc on which parity data is recorded (odn = 4).
It is an access frequency management table for n = 0, 1, 2, 3). Each access frequency management table 301, 30
In Nos. 2, 303, and 304, AP is the access frequency for each block, and AF is the access frequency ranking between blocks in the same optical disc. Here, the access frequency AP is represented by an integer from 0 to 15, and the larger the number, the higher the access frequency. The access frequency rank AF is represented by an integer from 0 to 3, and the smaller the number, the higher the access frequency. The access frequency ranking AF is determined by comparing the access frequency AP between the same address blocks of each optical disc.

Next, the operation of exchanging data between optical disks according to the access frequency will be described.

As shown in FIG. 4, in the present embodiment, the data (for example, D02 and D03) between the optical disks is exchanged between the same block addresses. That is, the data is exchanged within a range in which only the arrangement is changed without changing the combination of the block data in one RAID group. Therefore, it is not necessary to rewrite the parity data P0.

A specific example of the operation will be described below. In this example, the data of the optical disks in the same RAID group are replaced in the order of access frequency. For example, in FIG.
As shown in (a), before replacement, 501, 502, 5
It is assumed that there are optical disks odn = 0 to 3 of the same RAID group having the access frequency distributions shown in the access frequency management tables 03 and 504. When the access frequencies of these optical disks are replaced in the order of the set access frequency order AAF shown in FIG. For (obn), data is exchanged between the same blocks so that the access frequency rank AF becomes equal to the set access frequency rank AAF. As a result, 506,
The contents are updated in the access frequency management tables 507, 508, and 509.

FIG. 6 shows a flowchart of this data exchange process. This flowchart shows one identical block No. It shows the process of exchanging data between
This process applies to all block numbers. Is done about. Here, Dodn is optical disk (odn) data, AFod
n is the access frequency rank of Dodn, AAFodn is the access frequency rank of the optical disk (odn) after replacement, and HD is a hard disk corresponding to the auxiliary storage unit 108 in FIG. However, the auxiliary storage unit 108 is not limited to a hard disk, but may be a memory or any other medium capable of storing data. In addition, → indicates the movement of data, and in the case of an optical disc, the block number being processed. Read / write to.

Next, another embodiment will be described.

When one file data is written on one optical disk in a concentrated manner, as described in the previous embodiment, the block No. If the data is exchanged for each of them, there is a possibility that the file data will be scattered over a plurality of optical disks.

Therefore, as shown in FIG. 7, in the four optical disks odn = 0, 1, 2, 3 for storing file data, the start addresses of the respective file data stored in the individual optical disks are all the other three. The storage position of the file data is controlled so as to coincide with the start address or unused portion of the file data stored on one optical disk. Therefore, an unused area is provided between the end of one file data stored on the optical disk and the beginning of the next file data, if necessary. The numbers in the circles in the figure indicate the order in which the data files are stored. As described above, in the present embodiment, the size of the file data stored in each optical disk is as small as possible so that the start address of each file data stored in another optical disk matches the start address or the unused portion of the file data. Equal file data is collected and stored at the same position on each optical disk as much as possible.

When the file data is rewritten, the file data is re-stored in a storage location matching the size, as shown by → 'in FIG.

By storing the file data in this manner and exchanging the data by exchanging the data at the time of exchanging the data in units of the sections delimited by (a), (b) and (c) of FIG. 7, it is possible to prevent the files from being dispersed. .

Of course, also in this case, since only the data is exchanged between the same block addresses, it is not necessary to rewrite the parity data, and the data can be exchanged efficiently as described above.

In this embodiment, the case where the access frequency of each optical disk is concentrated has been described, but the present invention is not limited to such an apparatus, and the access frequency of the data stored in each optical disk is uniform. It may be applied to an apparatus for exchanging data so that

In this embodiment, a RAID system using an optical disk is taken as an example, but the present invention can be applied to a RAID system using another storage medium such as a hard disk.

[0031]

As described above, according to the present invention, by exchanging file data in block units among a plurality of first storage media at the same block address of each first storage media, No change is made to the block data of the RAID group. Therefore, it is not necessary to regenerate the parity data by this block data exchange, and the data exchange processing speed can be improved.

Further, when individual file data are concentrated and stored in one first storage medium, the start address of each file data stored in each first storage medium is set to the other first storage medium. By aligning with the start address or the unused portion of the file data stored in, it is possible to prevent one file data from being distributed to a plurality of storage media due to data exchange.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of a RAID system according to an embodiment of the present invention.

FIG. 2 is a diagram relating to management of file data.

FIG. 3 is a diagram related to management of access frequency of file data.

FIG. 4 is a diagram showing an example of exchanging data between optical disks in the present embodiment.

FIG. 5 is a diagram illustrating an example of a data replacement process according to an access frequency.

FIG. 6 is a flowchart showing a procedure of data replacement processing.

FIG. 7 is a diagram for explaining an embodiment in which individual file data are concentrated and stored in one first storage medium.

FIG. 8 is a diagram for explaining the basic concept of RAID.

FIG. 9 is a diagram showing exchange of data between conventional optical disks.

[Explanation of symbols]

101 ... Host computer, 102-106 ... Optical disk drive, 107 ... Optical disk, 108 ... Auxiliary storage section, 109 ... DMA, 110 ... Data generation section, 111 ...
Management unit, 201 ... File data management table, 30
1, 302, 303, 304 ... Access frequency management table.

Claims (2)

[Claims] 1. An information storage device in which file data is stored in blocks in a plurality of first storage media, and parity data for the same block of data in each first storage medium is stored in a second storage medium. The information storage device is characterized in that the file data of the block unit is exchanged between the plurality of first storage media between the same block addresses of the respective first storage media. 2. The information storage device according to claim 1, wherein, when individual file data are centrally stored in one first storage medium, each of the file data stored in each first storage medium is stored. An information storage device, characterized in that a start address and a file size are matched with a start address and a file size or an unused portion of file data stored in another first storage medium.