JP3181466B2 - Information storage system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information storage system comprising a disk array device having a parity disk.

[0002]

2. Description of the Related Art Conventionally, as a device for realizing high-speed data transfer and restoring data even if one disk drive or the like fails, a RAID (Re-
dundantArrays of Inexpensive Disks). RAID is, in other words, a disk array. Generally, a disk array is provided with a plurality of data storage disks and a parity disk for storing parity information of data stored in each data storage disk. In this case, as an operation for obtaining parity information, generally, an exclusive OR (EXCLUS
IVE-OR) is used. For this reason, even if one of the disk drives fails, the data stored in the failed disk drive is repaired based on the parity information on the parity disk and the data stored in the other disk drives. be able to.

[0003] As a system constituting a RAID,
"Distributed file system and storage device" (Japanese Patent Laid-Open No. 5-3)
24579). According to this system, a plurality of data disk drives are connected on a network via their own computers, and one of the data disk drives stores data stored in another data disk drive. Is stored. Here, if only one data disk drive has failed due to a disaster or the like, lost data can be recovered.

Further, an example of the configuration of a conventional basic disk array will be described with reference to FIG. In FIG.
Reference numeral 14 denotes an information processing device, which has a plurality of parity groups 141, 142,..., 14k. Each of these parity groups includes a plurality of optical disks, and forms a disk array. That is, the parity group 141 includes data storage disks 141a and 141b and a parity data storage disk 141c. Similarly, the parity group 142 includes the data storage disks 142c.
a, 142b and parity data storage disk 14
2c, and the parity group 14k includes data storage disks 14ka and 14kb and a parity data storage disk 14kc.

FIG. 15 shows the information processing apparatus 1 shown in FIG.
FIG. 2 is a block diagram showing an internal configuration of the device. In FIG.
Optical disks 141a to 14kc are connected to the bus 150. For example, when the data of any one of these optical disks is rewritten, the management unit 151 derives the parity data via the memory 152 and the EX-OR generation unit 153 in accordance with the rewriting, Store on the corresponding parity disk.

[0006]

As described above, conventionally,
In one information storage device, a method of storing redundant data relating to a plurality of media (such as disk drives) in another medium in the information storage device has been adopted.

However, in such an information storage device, for example, a failure, damage, theft, etc. occurs for some reason, such as a local disaster, an accident, an incident, etc., and a media group that matches and configures data redundancy. If two or more of the media become unreadable at the same time, there is a problem that data stored on the unreadable medium cannot be restored.

Also, in a system in which a plurality of information storage devices are distributed and arranged as a whole and can communicate data with each other, if the communication means between the information storage devices fails, communication becomes impossible. was there.

The present invention has been made in view of the above circumstances, and even if two or more disks become unreadable in an information storage device having a plurality of disks or the like, the data cannot be read. An object of the present invention is to provide an information storage system capable of restoring data and restoring data stored in an uncommunicable information storage device even when a communication means between the information storage devices fails. And

[0010]

According to the present invention, there is provided an information storage system which stores data managed by a local address.
A plurality of data disks for storing
At least one path for storing corresponding parity data;
And the difference disk and parity data
One or more first memories for storing;
Address indicating the correspondence between the address and the global address.
The dress management table and the first control means
And the first information storage device having the
Glow storing global parity data
Valparity disk and the required number of difference data
A second memory, and a local address and a global address.
Second address management table showing correspondence with dress
And a second information processor having at least a second control unit.
And the first device.
Dress management table and second address management table
Aligns global parity groups
An information storage system, comprising: the first information storage device;
The first control means may further include new data from outside.
When written on existing data on the data disc,
Reading the existing data from the data disk,
Exclusive OR operation between existing data and new data
Generate difference data and store it in the first memory
Means and existing parity data corresponding to the existing data
Is read from the parity disk, and the existing parity is read.
The exclusive OR operation of the difference data and the difference data
Generates new parity data and stores it in the first memory.
Means for storing the new data in the parity disk
To the existing data storage location of the
The parity data of the parity disk.
Means for writing to the storage location of the security data;
The difference data stored in the second information storage device
Means for transferring the information to the front of the second information storage device.
The second control means may further comprise the first information storage device.
The difference data sent from the second memory is stored in the second memory.
As well as existing global data corresponding to the differential data.
Data from the global parity disk
From the differential data and the global parity
Exclusive OR operation with data creates a new global
And the existing global parity data.
Data writing means for writing data to the storage location of the data .

[0011]

[0012]

With the above arrangement, even if two or more disks become unreadable in an information storage device having a plurality of disks or the like, restoration of unreadable data becomes impossible. it can.

[0013]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an information storage system according to one embodiment of the present invention. In the figure, 1
Is a global parity group, and the global parity group 1 is composed of a plurality of, for example, three information storage devices S1, S2, and S3. The information storage devices can mutually transmit and receive information by communication. Further, the information storage device S1 includes a plurality of terminals T1.
1, T12, and T13, and data can be rewritten and read from the information storage device S1 from any terminal. Similarly, the information storage device S2 includes a plurality of terminals T21, T22, T23.
Can be communicated with, from any terminal,
The data in the information storage device S2 can be rewritten and read.

The information storage device S1 is provided with a plurality of optical disks. The plurality of optical disks are divided into groups, and each optical disk is arranged so as to belong to one of the groups. Each group is defined as a local parity group S11, S12,..., S1m. The local parity group S11 includes a local parity disk S11a and data disks S11b and S11c.
have. Similarly, the local parity group S1
2 has a local parity disk S12a and data disks S12b and S12c, and the local parity group S1m includes a local parity disk S12m.
1ma and data disks S1mb, S1mc. The grouping of the optical disks is managed by a table in the information storage device S1 as described later.

Similarly, the information storage device S2 includes a plurality of optical disks. The plurality of optical disks are divided into groups, and each optical disk is arranged so as to belong to one of the groups. Let each group be a local parity group S21, S22, S2m. The local parity group S21 includes a local parity disk S21a and data disks S21b, S21c.
have. Similarly, the local parity group S2
2 has a local parity disk S22a and data disks S22b and S22c, and the local parity group S2m is a local parity disk S22m.
2ma and data disks S2mb, S2mc. The grouping of the optical disks is managed by a table in the information storage device S2 as described later.

As described above, each of the information storage devices S1 and S2 has an RAI having a plurality of disks.
D, and in this embodiment, the RAID in this case is particularly called a local RAID.

Each of the local parity disks in the information storage devices S1 and S2 stores parity data relating to data stored in a data disk in a local parity group to which the local storage belongs. The parity data at this time is data obtained by performing an exclusive OR operation on the data of each data disk. For example, in the information storage device S2 , the parity data of the local parity disk S21a is data obtained by performing an exclusive OR operation on the data of the data disks S21b and S21c. Therefore, even if the data on the data disk S21b becomes unreadable, the data on the data disk S21b is calculated by performing an exclusive OR operation on the data on the data disk S21c and the parity data on the local parity disk S21a. Can be derived.

The information storage device S3 includes a plurality of optical disks,
That is, the global parity disks S3a, S3b, S3
, S3n. Each global parity disk constitutes a group different from the above-mentioned local parity group together with the optical disks arbitrarily selected in the information storage device S1 and the information storage device S2. For example, the global parity disk S3
b constitutes a group different from the local parity group together with the optical disks S12a and S22c (not shown). The grouping of the optical disks in this case is managed by a table in the information storage devices S1 and S2 as described later.

Each global parity disk in the information storage device S3 stores parity data related to data stored on an optical disk in a group different from the local parity group to which the global parity disk belongs. In particular, this parity data is called global parity data.
The global parity data is stored in the information storage device S described above.
1 and the parity data in the information storage device S2, the data obtained by performing an exclusive OR operation on the data of each optical disk in the same group. For example, the global parity data of the global parity disk S3b is data obtained by performing an exclusive OR operation on the data of the optical disks S12a and S22c. Therefore, even if the data on the optical disk S12a becomes unreadable , the data on the optical disk S12a can be derived by performing an exclusive OR operation on the data on the optical disk S22c and the global parity data on the global parity disk S3b. .

FIG. 2 is a block diagram showing the internal configuration of the information storage device S1. In FIG. 2, an arbitrary terminal T11
Sends an access request to the terminal communication unit 21 in the information storage device S1 to write data on the optical disk or the like by data communication. Terminal communication unit 2
When receiving a data write instruction from the terminal T11, the device 1 notifies the management unit 22 of the instruction content via the bus and stores the data to be written in its own memory (not shown). The management unit 22 includes a table and a CPU (centra) for performing address management described later.
l processing unit) 221. This CPU
221 manages various processes performed in the information storage device S1. Here, the CPU 221 sends information of the new data to be written and the information of the existing data written in the expected write location (on any data disk) based on the above notification to the difference data generation unit 23. The difference data generation unit 23 receives the existing data and the new data sent by the CPU 221, performs an exclusive OR operation, and generates an operation result. The calculation result at this time is particularly called “difference data”. This difference data will be used later for deriving new parity data and deriving new global parity data. The CPU 221 stores the difference data generated by the difference data generation unit 23 at a predetermined position in the memory 24.

Further, the CPU 221 reads an existing parity written on a parity disk (new parity data writing location) corresponding to the difference data and the new data writing location stored in the memory 24. The data information is sent to the difference data generation unit 23.
The difference data generator 23 receives the difference data and the existing parity data sent by the CPU 221 and performs an exclusive OR operation to generate new parity data. The CPU 221 stores the new parity data generated by the difference data generation unit 23 at a predetermined position in the memory 24.

The CPU 221 further includes a terminal communication unit 21.
Is written to the planned write location (on the data disk), and the new parity data stored in the memory 24 is written to the planned write location (on the parity disk).

According to the above-described configuration, it can be said that the function of rewriting the corresponding parity data in accordance with the rewriting of arbitrary data is provided. Next, the CPU 221
Transfers the difference data stored in the memory 24 to the difference data storage unit 25 simultaneously with the writing of the new data and the new parity data. The difference data accumulation unit 25 accumulates the difference data sent from the memory 24 each time.
In accordance with the instruction from the PU 221, the accumulated amount of the difference data is sent to the difference data communication unit 26. The difference data communication unit 26
The difference data stored from the difference data storage unit 25 is transmitted to the information storage device S3. An address indicating a global parity disk to be rewritten is added to each difference data before the difference data is transmitted by referring to an address management table described later.

Although the configuration inside the information storage device S1 has been described so far, the configuration inside the information storage device S2 has the same configuration as described above. FIG. 3 shows the information storage device S
FIG. 3 is a block diagram showing the internal configuration of FIG. In FIG.
The difference data receiving unit 31 stores the information storage device S1 (or S
When receiving the accumulated amount of difference data sent from 2), it notifies the management unit 32 of the received content and transfers the accumulated amount of difference data to the difference data accumulation unit 33 via the bus.
The management unit 32 has a CPU 321. This CPU
321 manages various processes performed in the information storage device S3. On the other hand, the difference data accumulation unit 33 accumulates the accumulated difference data every time it receives the accumulated amount of the difference data transferred from the difference data reception unit 31, and accumulates the difference data until it reaches a predetermined amount.

When the difference data has been accumulated in the difference data accumulation unit 33 to a predetermined amount,
The accumulated difference data and the information of the existing global parity data written on the corresponding global parity disk (new global parity data writing place) are sent to the redundant data generation unit 34. The redundant data generation unit 34 receives the accumulated difference data sent from the CPU 321 and the existing global parity data, performs an exclusive OR operation, and generates new global parity data.
The CPU 321 stores the new global parity data generated by the redundant data generation unit 34 at a predetermined position in the memory 35. In addition, the CPU 321
Overwrites the new global parity data stored in the storage location (on the global parity disk) with the planned write.

According to the above configuration, it can be said that the function of rewriting the global parity data in the corresponding information storage device S3 in accordance with the rewriting of arbitrary data in the information storage device S1 or S2. .

Next, the information storage devices S1, S2, and S3
RAID management in each management unit and a management table used at that time will be described. FIG. 4 is a diagram illustrating RAID management using a management table in each management unit.
4A illustrates RAID management related to the information storage device S1, FIG. 4B illustrates RAID management related to the information storage device S2, and FIG.
(3) shows the RAID management for the information storage device S3.

In FIG. 4A, the management unit 22a
It has an address management table 41A and a local parity group management table (hereinafter, referred to as an LPG management table) 41B, and manages addresses of all the optical disks S11a to S1mc provided in the information storage device S1. Local addresses (LA) are given to the optical disks S11a to S1mc, respectively, and the correspondence between these local addresses and global addresses (GA) 0 to n is defined in the address management table 41A. In the LPG management table, a local address and a local parity group (L
PG) The correspondence with S11 to S1m is defined. In the LPG management table, the numbers indicating the LPGs are simply represented by 0 to m.

Similarly, in (2) of FIG.
2b has an address management table 42A and an LPG management table 42B, and performs address management of all the optical disks S21a to S2mc provided in the information storage device S2. Local addresses are given to the optical disks S21a to S2mc, respectively, and the correspondence between these local addresses and global addresses 0 to n is defined in the address management table 42A.
In the LPG management table, the correspondence between local addresses and local parity groups S21 to S2m is defined.

In (3) of FIG. 4, the management unit 32 has an address management table 43A and manages the addresses of the optical disks S3a to S3n provided in the information storage device S3. Local addresses are given to the optical disks S3a to S3n, respectively, and the correspondence between these local addresses and global addresses 0 to n is defined in the address management table 43A.

The above address management tables 41A, 42
In A and 43A, a flag (GPF) is attached to each local address. This flag indicates the location where the corresponding global parity data should be stored.

Here, in each of (1), (2) and (3) of FIG. 4, for example, the case where the global address is 1 is considered. In each address management table, the hatched portions represent the local address (LA) corresponding to the global address (GA) = 1 and the optical disk. In (1) of FIG. 4, the global address (G
A) = 1 corresponds to the local address (LA) = 3 and the optical disk S12a. In (2) of FIG. 4, the global address (GA) = 1, the local address (LA) = 5, and the optical disk S22c correspond. Further, in (3) of FIG. 4, when the global address (GA) = 1, the local address (L
A) = 0, corresponding to the optical disc S3a. Therefore,
In this case, the optical disks S12a and S2 indicated by oblique lines
According to the relationship between 2c and S3a, it can be said that the parity of the global parity group is matched.

In this embodiment, all the global parity data for the global parity group 1 is stored in the information storage device S3. However, the present invention is not limited to this, and the storage location of the global parity data can be freely set for each global address. As information storage devices S1, S
2, one can be selected from S3.

Each of the information storage devices S1, S2, S3
Have a plurality of optical disks, but only the information storage devices S1 and S2 constitute a local parity group. In the information storage device S3, since no local parity group is configured,
No LPG management table is provided. Information storage device S1
Is managed by the LPG management table 41B, and the local parity group in the information storage device S2 is managed by the LPG management table 42B.
B. In the LPG management table 41B of FIG. 4A, focusing on, for example, the local parity group (LPG) = S12, the corresponding local address (LA) = 3, 4, 5, that is, the optical disks S12a, S12b, and S12c. , A local parity group is formed, and parity matching is achieved.

In this embodiment, the optical disks constituting the local parity group are selected in the order of the optical disk numbers. However, the present invention is not limited to this, and the optical disks can be freely selected. As for the parity disk for storing parity data, one disk can be freely selected from the optical disks constituting the local parity group. Also, (1), (2),
In (3), the basic unit handled in the RAID management is one optical disk, but any unit having a uniform data size within a group such as a global parity group and a local parity group may be used. I do not care.

FIG. 5 shows RAID management when the basic unit is a block of an optical disk. (1)-of FIG.
(3) shows a case where the optical discs in (1) to (3) of FIG. 4 are replaced with blocks on the optical disc.

In FIG. 5A, the management unit 22a
Address management table 51A and LPG management table 5
1B, and manages addresses of blocks on the optical disks S11a to S1mc provided in the information storage device S1. Each block on the optical disks S11a to S1mc is given a local address (LA), and the address management table 51A defines the correspondence between the local address and the global addresses (GA) 0 to n. In the LPG management table, the local address and the local parity group S
The correspondence with 11 to S1m is defined. At this time, in each management table, the local address (LA) is
It is represented by a combination of an optical disk number and its block number.

Similarly, in (2) of FIG.
2b has an address management table 52A and an LPG management table 52B, and manages addresses of blocks on the optical disks S21a to S2mc provided in the information storage device S2. Each block on the optical discs S21a to S2mc is given a local address (LA). In the address management table 52A, a local address and a global address (GA) 0 are assigned.
To n are defined. In the LPG management table, the correspondence between local addresses and local parity groups S21 to S2m is defined.

In (3) of FIG. 5, the management unit 32 has an address management table 53A, and manages addresses of blocks on the optical discs S3a to S3n provided in the information storage device S3. A local address is given to each block on the optical disks S3a to S3n, and a correspondence between the local address and the global addresses 0 to n is defined in the address management table 53A.

The above address management tables 51A, 52
A and 53A, the local address is represented by a combination of an optical disk number and a block number on the optical disk, and a flag (GP
F) is attached.

Here, in each of (1), (2) and (3) in FIG. 5, for example, the case where the global address is 1 is considered. In each address management table, the hatched portions represent the local address (LA) and the block corresponding to the global address (GA) = 1.
In (1) of FIG. 5, the global address (GA)
= 1, the local address (LA) = 0-5 corresponds to the block 51a. In (2) of FIG. 5, the global address (GA) = 1 corresponds to the local address (LA) = 2-4 and the block 52a. Further, in (3) of FIG. 5, when the global address (GA) = 1, the local address (L
A) = 0, corresponding to block 53a. Therefore, in this case, the blocks 51a, 52a, 5
By 3a, it can be said that the parity of the global parity group 1 is matched.

In FIG. 4 and FIG. 5, three information storage devices constituting the global parity group 1 and three optical disks or three blocks constituting the local parity group are used. The number is not limited, and may be any number. Also,
In each of the information storage devices S1 to S3, even if the configuration of each local parity group is different, the size (storage capacity) of each group address is unified,
It is sufficient that the group addresses 0 to n are associated with each other. However, in this case, it must be managed by the corresponding address management table and LPG management table. Further, in this embodiment, an optical disk is used as a storage medium, but any type of data can be used as long as it can store data.

In the RAID management described with reference to FIGS. 4 and 5, the reliability of the management table is important. If the address management table or the LPG management table in the management unit is lost, it is impossible to know in what combination the storage data in the information storage device participates in each management table. Therefore, each management table must be stored in an extremely reliable state. Therefore, RAID management with higher reliability will be described.

FIG. 6 shows a RAID with further improved reliability.
It is a figure showing management. (1) and (2) in FIG. 6 have basically the same configuration as (1) and (2) in FIG. (1) of FIG. 6 has the same configuration as (1) of FIG. 4 except that the flag (GPF) does not exist in the address management table 61A. Further, (2) in FIG. 6 has the same configuration as (2) in FIG. 4 except that the flag (GPF) does not exist in the address management table 62A. FIG. 6C has an address management device 63A and an LPG management device 63B.
It does not manage the correspondence with the addresses 3a to A3n. The address management device 63A has the result of the exclusive OR of the data in the address management table 61A and the data in 62A, that is, the parity data of 61A and 62A. For this reason, the address management table 61A or 6
This can be restored even if 2A is lost. LPG
The management device 63B calculates the result of the exclusive OR of the data in the LPG management table 61B and the data in 62B, that is,
It has 61B and 62B parity data. For this reason,
Even if the address management table 61B or 62B is lost, it can be restored. As described above, the parity information on the storage data in the optical discs S11a to S1mc and S21a to S2mc is stored in the optical discs S3a to S3mc.
It is stored in S3n. Therefore, the parity information is used as global parity data in the address management device 63 in the management unit 32 in the information storage device S3.
A, LPG management device 63B, and optical disks S3a to S3S
3n.

In the information storage device S3, like the information storage devices S1 and S2, the optical disks S3a to S3
It does not have a table for managing the correspondence with the address of n . In this case, there is a method of associating a global address with each optical disk using another extremely high-performance storage device (not shown), but there is also a method of physically associating the two with some kind of means. At this time, a table for associating the two is not required.

As described above, by storing each management table with redundancy in the direction of the global parity group and storing the parity information in the information storage device S3, the reliability becomes extremely high.

Next, the operation when parity matching is performed in the above embodiment will be described. FIG. 7 and FIG. 8 are diagrams showing a processing procedure for performing parity matching. FIG. 7 shows a processing procedure in the information storage device S1 or S2, and FIG. 8 shows a processing procedure in the information storage device S3.

In (1) of FIG. 7, the information storage device S1
Alternatively, when S2 responds to an access request from an arbitrary terminal, new data ND is received in the terminal communication unit. By calculating an exclusive OR (EX-OR) between the data ND and the existing data D written at the place where the data ND is to be written (on the data disk 71), difference data ΔD is generated. The difference data ΔD is stored in the memory 24. In (2), the exclusive OR (EX-OR) of ΔD stored in the memory 24 and the existing parity data P at the location corresponding to the local address of the location to be rewritten to the data ND (on the parity disk 72) ) To generate new parity data NP, and store this parity data NP in the memory 24. In (3), the data ND stored in the terminal communication unit 21 is written to the planned write location, and the parity data NP stored in the memory 24 is written to the planned write location. At the same time, the difference data ΔD stored in the memory 24 is transferred to the difference data storage unit 25. Further, the difference data ΔD0 to ΔDn sent to the difference data storage unit 25 each time are stored in the difference data storage unit 25, and when the difference data is stored to a predetermined amount, the difference data ΔDs0 (the difference data ΔD
0 to ΔDn) are transmitted from the difference data transmitting unit 26 to the information storage device S3. At the time of transmission, a corresponding global address is added for each difference data.

In (1) of FIG. 8, the information storage device S1
Alternatively, the difference data ΔDs0 sent from S2 is received by the difference data receiving unit 31 in the information storage device S3. The difference data ΔDs0 received by the difference data receiving unit 31 is transferred to the difference data storage unit 33. Difference data storage unit 33
The difference data ΔDS0 and the like sent each time are stored together with the already stored difference data ΔDS1 (a plurality of difference data are grouped together like the difference data ΔDS0). In (2), when the difference data is accumulated in the difference data accumulation unit 33 until the difference data reaches a predetermined amount, the accumulated difference data is stored in a location (on the global parity disk 81) corresponding to the difference data. By calculating the exclusive OR with the current global parity data GP, the new global parity data N is calculated.
GP is generated, and the global parity data NGP is stored in the memory 35. In (3), the global parity data NGP stored in the memory 35 is overwritten on the location where the existing global parity data GP is written.

FIG. 9 shows how the relationship between the transfer of difference data between information storage devices and parity matching changes over time. In FIG. 9, TS1, TS2, and TS3 indicate time axes for the information storage devices S1, S2, and S3, respectively, and each double line indicates parity matching at that time. As described above, the difference data is transmitted from the information storage device S1 or S2 to the information storage device S3.

For example, at time 91a of TS1, TS
When the differential data is transmitted from 1 to TS3, the time 91
Parity matching is achieved by double lines 9A1 and 9A2 connecting a, 92a and 93a. Next, at time 9 of TS2
When the difference data is transmitted from TS2 to TS3 at 3b, the above-mentioned double line is changed to the time points 91a, 92b, 93b.
Move to the double lines 9B1 and 9B2 connecting. Further, when the differential data is transmitted from TS1 to TS3 at the time point 91b of TS1, the above-mentioned double line becomes the time points 91b, 92c,
Move to the double lines 9C1 and 9C2 connecting 93b. After that,
The same operation is repeated over time. As described above, parity consistency shifts over time.

When transferring the difference data, data indicating the date and time when the difference data was generated is also transferred. FIG. 10 shows a table for managing the date and time of difference data generation in each information storage device. In this table, the date and time when the difference data was generated is recorded for each information storage device. By doing so, it is possible to clarify at what point the stored data is restored when restoring the case where data reading becomes impossible.

Next, a case where a new information storage device is added or a case where a new optical disk is added will be described. FIG. 11 is a diagram illustrating a state where a new information storage device and the like are added. In FIG. 11, global parity group 1 includes existing information storage devices S1, S2, and S3.
In this global parity group 1, a new information storage device S4 can be made to participate. Further, a new optical disk 11a can be added to the information storage device S4.

FIG. 12 is a diagram showing RAID management in consideration of addition of a new optical disk. 12, the management unit 22c in the information storage device S4 has an address management table 12A, an LPG management table 12B, and a plurality of local parity groups S41, S42,..., S4m. The local parity group S41 includes optical disks S41a to S41c, and the local parity group S42 includes only optical disks S42a and S42b. Note that the local parity group S4m does not include an optical disk. Within the LPG management table 12B, an empty mark is added to each address indicating an optical disk. As the value of the empty mark, “0” is used for an optical disk that does not physically exist, and the existing optical disks S41a to S41c, S42a, S42
"1" is used for b.

Thus, even if an optical disk does not actually exist, an optical disk that does not physically exist can participate in the parity group. Therefore, a new optical disk can be added at any time by freely allocating a new optical disk to a location having an empty mark.

However, when a new information storage device S4 is added, the time management table for generating difference data shown in FIG.
It must be changed so that the time management of the information storage device S4 newly joined can be performed. FIG. 13 shows a time management table after the change.

As described above, according to the present invention, even when a plurality of optical disks fail simultaneously, the optical disk can always be recovered. Further, for example, one of three information storage devices
Even if one of the units is completely disabled for some reason, it can be recovered using the other two information storage devices.

[0058]

As described above in detail, according to the present invention, local disasters are managed by managing storage data by a storage system having redundancy among storage media of a plurality of distributed information storage devices. The reliability of stored data for accidents, incidents, and the like can be improved.

Further, when the communication means between the distributed information storage devices breaks down and communication becomes impossible, the storage of the information storage device which cannot be communicated from other communicable information storage devices having redundancy is performed. Data can also be generated.

Further, when there are a plurality of information storage devices which store data with redundancy among a plurality of local storage devices, a storage system having redundancy between the information storage devices is used. By managing the stored data and making the stored data redundant in a hierarchical manner, the reliability can be further improved. In addition, it is not limited to two levels,
It can be applied to further multi-layers. In this case, the reliability can be further improved by increasing the number of layers to three layers from two layers and four layers from three layers. Further, the storage medium is not limited to an optical disk, and can be applied to other hard disks and the like.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an information storage system according to an embodiment of the present invention.

FIG. 2 is an internal configuration diagram of information storage devices S1 and S2 according to the embodiment.

FIG. 3 is an internal configuration diagram of an information storage device S3 according to the embodiment.

FIG. 4 is a diagram showing RAID management for each optical disc.

FIG. 5 is a diagram showing RAID management for each block.

FIG. 6 is a diagram showing RAID management with improved reliability.

FIG. 7 is an exemplary view showing a processing procedure of the information storage devices S1 and S2 according to the embodiment.

FIG. 8 is an exemplary view showing a processing procedure of the information storage device S3 according to the embodiment.

FIG. 9 is a diagram showing a transition of parity matching with respect to a temporal change.

FIG. 10 is a diagram showing management of date and time of difference data generation.

FIG. 11 is a diagram showing addition of a new information storage device or optical disk.

FIG. 12 is a RAID in consideration of addition of a new optical disc;
The figure which shows management.

FIG. 13 is a diagram showing management of date and time when a new information storage device participates.

FIG. 14 is a configuration diagram of a conventional information storage device.

FIG. 15 is a diagram showing the internal configuration of a conventional information storage device.

[Explanation of symbols]

1: Global parity group, S1 to S4, 14 ...
Information storage device, S11 to S1m, S21 to S2m, 14
1 to 14k: Local parity group, S11a to S11
1mc, S21a to S2mc, S3a to S3n, 11
a, S41a to S42b, 71, 72, 81 ... optical disk, T11 to T13, T21 to T23 ... terminal, 21 ... terminal communication unit, 22, 32, 22a, 22b, 151 ... management unit, 221, 321 ... CPU, 23 ... difference data generation unit, 24, 35, 152 ... memory, 25, 33 ... difference data storage unit, 26 ... difference data communication unit, 31 ... difference data reception unit, 34 ... redundant data generation unit, 41A to 43A,
51A to 53A, 61A, 62A, 12a ... address management tables, 41B, 42B, 51B, 52B, 12b
... Local parity group management table, 51a, 5
2a, 53a ... block, 63A ... address management device,
63B: Local parity group management device, TS1
TS3: time axis, 9A1, 9A2, 9B1, 9B2, 9
C1, 9C2: parity matching, 91a, 91b, 92a
~ 92c, 93a, 93b ... point, 153 ... EX-OR
Generation unit.

Claims (1)

(57) [Claims] 1. A data managed by a local address.
A plurality of data disks for storing data;
At least one for storing parity data corresponding to the data
Parity disk and differential and parity data
One or more first memories for storing data,
First showing the correspondence between addresses and global addresses
Address management table and the first control means
A first information storage device and a global parity managed by a local address
A global parity disk for storing data,
A second memory for storing a required number of difference data;
The second showing the correspondence between addresses and global addresses
Address management table and the second control means
At least configured, the the first address management table and the second address management table is information storage system matching the global parity group is taken, the first information by the second information processing apparatus also has a The first control means of the storage device further comprises:
In addition, new data from the outside
Data is written from the data disk to the existing
Read the existing data and read the existing data and the new data
The difference data is generated by an exclusive OR operation with
Means for storing in the first memory, and existing parity data corresponding to the existing data,
Read from the parity disk and the existing parity data
A new parameter is calculated by an exclusive OR operation of the
Generating data and storing the data in the first memory.
And the new data is transferred to the existing data of the parity disk.
The new parity data.
Storage of the existing parity data on the parity disk
Means for writing to a location; and storing the difference data stored in the first memory in the second memory.
Means for transferring the information to another information storage device, wherein the second control means of the second information storage device further comprises:
The difference data sent from the first information storage device.
Data in the second memory and the difference data.
The existing global parity data corresponding to the
Read from the global parity disk, and
Exclusive OR operation of the global parity data and the global parity data
Generate new global parity data by
How to write to the existing global parity data storage location
Information storage system characterized by having steps.