JP3248537B2 - Data distribution storage method, data distribution storage device, and data processing system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of distributing and storing valuable data in a recording device installed in different areas in a distributed manner , thereby securely storing the data .
The present invention relates to a data distribution storage device and a data processing system .

[0002]

2. Description of the Related Art Conventionally, data processed by a bank or the like may cause serious social problems if the data is damaged. Therefore, the reliability of the entire system is improved by regularly backing up data. I have. Further, in this backed-up data, not only one location but also a plurality of locations are duplicated and distributed and stored, so that in this type of data processing system, one location of the backup data is damaged. In addition, the system can be backed up using data in other places.

[0003]

However, when the backup data is redundantly stored as described above, there is a problem that the redundancy of the entire system increases and the amount of data to be stored increases accordingly. This redundancy is represented by the following equation. For example, when data is redundantly stored at two locations, the redundancy is 100%.

[0004]

(Equation 1)

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and it is possible to reduce the amount of data to be stored when processing data is stored in a distributed manner, and to reproduce even when a part of the processing data is missing. Yes , especially multiple that are part of the processing data
Data that can be reproduced even if the data is missing
Distributed storage method, data distributed storage device, and data processing system
The purpose is to provide a system.

[0006]

SUMMARY OF THE INVENTION The above object is achieved by the present invention as defined in claim 1.
In the invention, the processing data is processed by the main processor.
Is divided into predetermined units to generate a data group of unit data.
By the dividing step and the main processor
By arranging the unit data vertically, the processing data
Are arranged in the form of a matrix, and the unit which is continuous in the vertical direction is
Generates a parity data string as data parity data
The parity data generation step and the error in the transmission path.
The detection code is added, and the
The data group and the parity data string are divided into the predetermined units.
Each time it is transmitted to the corresponding one of multiple sub-processors
By the transmitting step and each of the sub-processors,
Error detection processing based on the error detection code in the transmission path
And if an error is detected, the main processor
A retransmission request step for sending a retransmission request to the
The data group and the parity are processed by the sub-processor.
Other parity data is generated for each predetermined unit of the
Generating another parity data, and
The data group and the data group are stored in an external storage device of the processor.
And for each of the predetermined units of the parity data string,
Storing together with the security data,
The processing data of the predetermined unit can be read from the external storage device.
If not, these reads are performed by the main processor.
The parity data corresponding to the data not to be
Unreadable data using other parity data
Reproducing a plurality of sub-programs.
Rosesas must be located in different regions
This is achieved by a data distributed storage method characterized by the following.

[0007]

According to a second aspect of the present invention, in addition to the parity data and the other parity data, the unreadable data is read using readable data corresponding to the parity data and the other parity data . It is characterized by performing reproduction. According to the third aspect of the present invention, the object is to divide processing data into predetermined units,
Main processor division means to generate data group of data
And arranging the unit data in the vertical direction,
Before arranging data in a matrix,
Parity data as parity data of the unit data
A parity data generator of the main processor for generating a column;
And an error detection code in the transmission path,
The data group and the parity data generated by the processor;
And a plurality of sub-processors for each of the predetermined units.
Transmission means for transmitting to a corresponding one of the
Perform error detection processing based on the error detection code in
If an error is detected, retransmission is required to the main processor.
Requesting, a retransmission requesting means of each of the sub-processors,
The predetermined unit of the data group and the parity data string;
Each of the sub-processes for generating other parity data for each
Other parity data generating means and each of the sub-processors.
The data group and the
For each of the predetermined units of the parity data string, the other parity
Storage means for storing together with the external data,
If the predetermined unit of processing data cannot be read from the storage device,
In this case, the paris corresponding to these unreadable data
Using the parity data and the other parity data
The main processor re-creates data that cannot be written or read.
Current means, and wherein the plurality of sub-processors are different
It is characterized by being arranged in each region
This is achieved by a data distributed storage device . Claim 4
The present invention has a configuration in which in addition to the parity data and the other parity data, reproduction of the unreadable data is performed by using readable data corresponding to the parity data and the other parity data. It is characterized by the following.

[0009]

According to the above arrangement, even if one of the data groups stored in a distributed manner cannot be obtained, the damaged data group can be reproduced using the other data groups and the parity data . Furthermore, the processing data can be recorded on different recording
Data included in the data group distributed and stored in
Even if data could not be read, multiple data that could not be read
Use corresponding parity data and other parity data
To reproduce multiple data that could not be read
Data can be obtained.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below in detail with reference to the accompanying drawings. Note that the embodiments described below are
Since it is a preferred specific example of the present invention, various technically preferred limitations are added, but the scope of the present invention is limited to the following description unless otherwise specified to limit the present invention.
It is not limited to these embodiments.

FIG. 1 is a block diagram showing one embodiment of a data processing system according to the present invention. In the figure, a data processing system 1 (data distributed storage device) serially transmits processing data of a main processor M via a bus 2, whereby N + M sub-processors S 1 to SN +
M through a data recording device (recording device).
And record them separately . Here, the secondary processors S1 to S1
The SN + Ms are located in different regions, respectively, so that the processors S1 to S1 can be located at one place due to a natural disaster such as an earthquake.
Even when SN + M stops operating, the remaining processors S1 to SN + M can operate.

That is, the main and sub processors M and S1
.About.SN + M are microprogram-controlled processors whose operations are controlled by microprocessors (MPUs) 6 and 8, respectively, and operate in synchronization with each other. Further, the main and sub processors M, S1 to
SN + M are data input registers (DI) 10
And 12, data can be input from the data bus 2 as necessary, and the parity code added to the data is input to a parity check circuit (PC) 14, 1
In the case of a bit error for which error correction is difficult by performing a parity check in step 6, a data retransmission request is transmitted to the bus 2.

In this case, the main and sub processors M, S1 to
The SN + M is obtained after the data is reproduced by correcting the bit error in the microprocessors 6 and 8 depending on the response from the transmission target obtained via the bus 2 and the bit error in the range in which the error can be corrected. Output to an external storage device or the like as necessary. At the time of this data output, the main and sub processors M and S1 to SN add predetermined parity data and record the data in the external storage device, whereby the data once stored can be reliably reproduced. ing.

Further, main and sub processors M, S1 to S
N + M are parity bit addition circuits (PG) 18, 20
The desired data can be transmitted to the bus 2 via the bus. At the time of this data transmission, the parity bit adding circuit (P
G) At 18 and 20 , desired parity data is added and output.

Thus, the primary and secondary processors M, S1
.. SN + M can input and output data via the bus 2. At this time, parity data is added to output data and bit correction can be performed. When storing in
Further, parity data is added and output.

In the data input / output, when the data of the main processor M is distributed and stored, the main processor M
(Dividing means, parity data generating means), as shown in FIG. 2, divides the original information data (processed data) to be transmitted into predetermined units, thereby dividing the data group as N first data groups. A ₁ (a ₁₁ , a ₁₂ ,..., A _1L ), A _{2
(A 21, a 22, ......} , a 2L), ......, A N (a N1,
a _N2 ,..., a _NL ) (division step).

Further, to satisfy the following expression, the main processor M sets each data group A ₁ (a ₁₁ , a ₁₂ ,.
_{a 1L), A 2 (a} 21, a 22, ......, a 2L), ......, A N
(A _N1 , a _N2 ,..., A _NL ) are arranged in the vertical direction, and parity data strings P ₁ (p ₁₁ , p ₁₂ ,..., P _1L ) and P ₂ as parity data continuous in the vertical direction are arranged. (P ₂₁ ,
_{p 22, ......, p 2L)} , ......, P M (p M1, p M2, ......,
p _ML ) (parity data generation step) .

[0018]

(Equation 2)

Here, H ₁ is a Reed-Solomon code parity check matrix of M rows (N + M) columns. The N + M data groups A ₁ (a ₁₁ , a ₁₂ ,
_{......, a 1L), A 2} (a 21, a 22, ......, a 2L), ...
.., A _N (a _N1 , a _N2 ,..., A _NL ) and parity data string P ₁ (p ₁₁ , p ₁₂ ,..., P _1L ), P ₂ (p ₂₁ , p _21
22 ,…, p _2L ),…, P _M (p _M1 , p _M2 ,…, p
_ML ), the main processor M sequentially sends out to the sub processors S1 to SN + M according to a predetermined protocol.

As a result, in the data processing system 1, valuable data that needs to be stored and parity data for error correction are distributed and recorded in recording devices in each region, thereby reducing the amount of data required for recording. It can be reduced. In the sub processors S1 to SN + M, the transmitted data group A ₁ (a ₁₁ , a ₁₂ ,.
_{a 1L), A 2 (a} 21, a 22, ......, a 2L), ......, A N
(A _N1 , a _N2 ,..., A _NL ) and the parity data string P _{1
(P 11, p 12, ......} , p 1L), P 2 (p 21, p 22, ...
, P _2L ),..., P _M (p _M1 , p _M2 ,..., P _ML ) are subjected to an error detection process added on the transmission line. If no error is detected here, a predetermined external that records in the storage device (a first storing step).

At this time, when an error is detected, a retransmission request is sent to the main processor M, and the transmitted data is subjected to error detection processing and then recorded in an external storage device. When data is stored in the external storage device, as shown in FIG. 3, the sub processors S1 to SN + M respectively transmit K bytes of parity data (q ₁₁ , q ₁₂ ,. q _1K ), (q ₂₁ , q ₂₂ ,..., q
_2K ), ..., ( _qN1 , _qN2 , ..., _qNK ), (qN _{+ 11} ,
q _{N + 12} , ..., q _{N + 1K} ), (q _{N + 21} , q _{N + 22} , ..., q
_{N + 2K} ),..., (QN _{+ M1} , qN _{+ M2} ,..., QN _{+ MK} ), and records this parity data (other parity data) together (second storage). Step)

That is, each of the sub processors S1 to SN + M
In the above, each of the following arithmetic processes is executed, and the transmitted data is stored in a predetermined external storage device.

[0023]

(Equation 3)

Here, H ₂ is a Reed-Solomon code parity check matrix of K rows (L + K) columns. As a result, in the data processing system 1, the data that needs to be stored is distributed and stored by adding parity data in the product code format as a whole. Therefore, in the data processing system 1, the sub-processor S
Even when the operations of 1 to SN + M are stopped, by executing the error correction processing based on the error flags obtained from the sub processors S1 to SN + M in the remaining districts, they are allocated to the stopped sub processors S1 to SN + M. Data can be reproduced. This will be specifically described with reference to FIG. 3. In the data processing system 1, data can be reproduced as follows. In the data processing system 1, for example, when one position data a ₁₂ has failed to read in the data group A _1, the data a ₂₂ data group A _2, ...... data a _N2 data group A _N
, P _M2 , the data a ₁₂ can be reproduced by the parity data p ₁₂ , p ₂₂ ,..., P _M2 . In the data processing system 1, for example if the data a ₁₂ and the parity data p ₂₂ a plurality of locations has failed to read in each data group A ₁ and the parity data sequence P ₂ are
Each data a _11, a _13, ......, the parity data q _11, q _12, based on _{a 1L,} ......, data a ₁₂ by q _1K
Can be reproduced, and the parity data p ₂₁ , p ₂₃ ,...
..., parity data based on the _{p 2L q N + 21, q} N + 22, ...
.., Q _{N + 2K} can reproduce the parity data p ₂₂ . Further, in the data processing system 1, after is reproduced parity data p ₂₂ of the parity data sequence P ₂ described above, it may be so as to reproduce the data a _12. Therefore, in the data processing system 1, not only the data of the data group A ₁ only, a case of lacking data for various combinations to reproduce also data, obtaining processed data comprised of these data be able to
(Reproduction step) .

In this way, when the product code is added in the form of a product code,
At the same time, error correction can be performed until the M sub-processors S1 to SN + M stop operating, whereby the amount of data required for storage can be reduced and desired data can be reliably stored. That is, each of the sub processors S1 to SN + M
When a data read command is sent from the main processor M, the data sent to the external storage device is sent according to a predetermined protocol.

At this time, each of the sub processors S1 to SN + M
Executes an error detection process based on the parity data added at the time of recording, and sends an error flag if the error cannot be corrected or if the data cannot be sent. In response, the main processor M sets the sub processors S1 to S
After fetching the data transmitted from N + M, error detection processing is executed based on the parity data on the transmission path added during transmission.

If an error is detected or an error flag is transmitted from the sub processors S1 to SN + M, the main processor M executes an error correction process based on the error detection result and the error flag. As a result, the data once distributed and stored is reproduced.

According to the above configuration, by adding parity data in a product code format to data that needs to be stored and distributing and storing the data, even if a part of the distributed and stored data is missing, the parity data can be added to the data. The data can be reproduced on the basis of the data, whereby the amount of data required for recording can be reduced, and valuable data can be reliably stored. In the above-described embodiment, the case where the error detection processing is executed and the error flag is transmitted when the data is transmitted from the sub-processors S1 to SN + M has been described. When the processors S1 to SN + M stop operating, an error flag may be manually transmitted.

[0029]

As described above, according to the present invention, parity data is generated by dividing processing data into a first data group having a preset unit length, and the first data group and the parity data string are generated. Are stored separately in different recording devices, so that, for example, even when a part of the recording device stops operating and lacks a part of the processing data, not only the missing processing data can be reproduced, but also the distributed processing can be performed. By storing the data, the amount of data can be reduced and valuable data can be securely stored , and the processed data can be distributed to different recording devices.
Reads multiple data included in the stored data group
Reproduce multiple data that could not be read without
A data distribution storage method that can obtain processed data,
Data storage system and data processing system.
Can be.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a data processing device according to the present invention.

FIG. 2 is a chart for explaining the data transmission operation.

FIG. 3 is a chart for explaining the data recording operation.

[Explanation of symbols]

Reference Signs List 1 data processing device 2 bus 6, 8 microprocessor M main processor (division unit, parity data generation unit) S1-SN + M sub-processor

Claims (4)

(57) [Claims] (1) Processing data is processed by a main processor.
Divide by a predetermined unit to generate a data group of unit data,
Dividing the unit data in the vertical direction by the main processor;
The processing data is arranged in a matrix
Then, the parity data of the unit data that is continuous in the vertical direction is
Parity data to generate a parity data string as data
Data generating step, and adding an error detection code in the transmission path,
The data group and the parity data string generated in
Each of the predetermined units corresponds to a plurality of sub-processors.
Transmitting to an object, and detecting an error in the transmission path by each of the sub-processors.
Performs error detection processing based on the code and detects errors.
Sends a retransmission request to the main processor
A retransmission request step, and the data group and the
Another parity data is provided for each of the predetermined units of the parity data string.
The parity data is generated by another parity data generating step and an external storage device of each of the sub processors.
Data group and the predetermined unit of the parity data string,
Storing together with the parity data
The predetermined unit of processing data is read from the external storage device.
If not, the main processor will read them.
The parity data corresponding to the invisible data and
The data that cannot be read using the other parity data.
And reproducing the data, wherein the plurality of sub-processors are respectively arranged in different regions.
A data distributed storage method characterized by being performed . 2. Reproducing the unreadable data using readable data corresponding to the parity data and the other parity data in addition to the parity data and the other parity data. 2. The data distributed storage method according to claim 1, wherein: 3. The processing data is divided into predetermined units, and
Main processor division means to generate data group of data
And arranging the unit data in the vertical direction, so that the processing data
Are arranged in a matrix, and the unit
Parity data string as parity data of position data
Generating the parity data of the main processor;
Stage and an error detection code in the transmission path,
The data group and the parity data string generated in
Each of the predetermined units corresponds to a plurality of sub-processors.
Transmission means for transmitting an error detection code based on an error detection code in the transmission path.
And if an error is detected, the main processor
A retransmission request to each of the sub-processors.
Requesting means, the predetermined unit of the data group and the parity data string.
Each of the sub-processes for generating other parity data for each
The parity data is generated by another parity data generating means of the processor and an external storage device of each of the sub processors.
Data group and the predetermined unit of the parity data string,
Storage means for storing together with the other parity data
The predetermined unit of processing data is read from the external storage device.
If the data cannot be read out,
The parity data and the other parity data
The main program that reproduces the unreadable data using
And a plurality of the sub-processors are respectively arranged in different regions.
Data distributed storage characterized by the configuration
apparatus. 4. A structure for reproducing the unreadable data using readable data corresponding to the parity data and the other parity data in addition to the parity data and the other parity data. 4. The data distributed storage device according to claim 3, wherein: