JP6053476B2 - calculator - Google Patents

Description

  The present invention relates to a computer constituting a database system, and more particularly to a technique for matching the contents of a plurality of databases.

  For example, in a monitoring control system for operating a power system, a large amount of data is sequentially recorded and managed as a database. In such a system, computers and networks are multiplexed for stable operation. For example, a plurality of databases are arranged in a distributed manner, and a stable operation can be performed by reflecting an update of one database to another database.

  A function for transferring update contents generated in a database on a certain computer to databases on a plurality of other computers via a network and matching the update contents in each database is called a replication function. For example, a database management system having a replication function based on a master / slave method is known. The replication function using the master / slave method accepts a data update request from a database of a specific one computer (master), and from the specific one computer (master) to another computer (slave). This is to notify update data to the database. In the case of this master / slave method, even if data update requests occur on multiple computers for the same record, the requests are processed in order in the master database, making it easy to match the database contents. It is.

  In recent years, database management systems having a data replication function called a multi-master method have become widespread. The multi-master method refers to a method in which a program running on each computer can directly update the database of the computer and the update result is reflected between the databases in a system including a plurality of databases.

  Comparing the multi-master method with the master-slave method, the master-slave method is limited to one specific computer that can update the contents of the database, and passes through this one specific computer (master). While the database of each computer is updated, each computer updates the database in the multi-master method. Therefore, in the multi-master method, the load of update processing can be distributed among the computers, and data writing can be accelerated compared to the master-slave method. In addition, because the load of update processing is distributed among computers, even if a specific database cannot be operated, it is possible to operate a database management system stably without affecting the update processing of other databases. There is an advantage that it is easy.

  However, in the multi-master method, since each computer updates the database at an arbitrary timing, data update can occur in a plurality of computers for the same record. Therefore, it is necessary to perform processing for matching the contents of the database of each computer among the computers.

  The following patent documents and the like are known as techniques for maintaining the consistency of the contents of a plurality of databases by the multi-master method. For example, Japanese Patent Laid-Open No. 3-256146 (Patent Document 1) discloses a primary database update sequence number, a secondary database update sequence number, and a record update time for a primary database and a secondary database configured to have the same contents. It describes a technology that integrates databases by holding them in database records and collating the update sequence numbers of each database. In Patent Document 1, records updated in parallel in each database are reprocessed by comparing record update times. Thereby, the order of database update is maintained. Japanese Patent Application Laid-Open No. 11-7403 (Patent Document 2) describes a technique for avoiding database inconsistency by comparing the execution time of replication with the update time of a record to be replicated.

JP-A-3-256146 Japanese Patent Application Laid-Open No. 11-7403

  When maintaining the consistency of the contents of a database using a plurality of parameters as in Patent Document 1, it is necessary to implement a complicated determination process using a plurality of parameters. In particular, the larger the database management system with a large number of records, the greater the processing load for matching the contents of the database. Further, as in Patent Document 2, in order to match the contents of the database based on time information such as the update time of a record, it is necessary to strictly match the time counted by each computer. Accurate time synchronization processing is required.

  Therefore, the present invention is a low-cost mechanism for maintaining the consistency of the contents of a database in a database system to which a multi-master system is applied in which a plurality of databases are distributed and updated from any database. The purpose is to provide technology realized in

  The computer according to the present invention provides a configuration of a database system that performs replication between a plurality of computers by a multi-master method. This computer stores an update value indicating an update status for each record held in the database. When updating a record, the computer calculates an updated value of the updated record based on the updated value corresponding to the record before the update and the priority value unique to the own computer. The updated value of the record after the update is a value corresponding to the number of record updates, and is a value unique to the own computer. When the record is updated by the own computer, the updated record and the calculated update value are transmitted to other computers constituting the database system. When the record is updated in another computer, the computer receives the updated record and the updated value from the other computer, and compares the received updated value with the updated value stored in the own computer. To replicate between multiple computers.

  According to the present invention, each computer can easily perform the process of matching the contents of a plurality of databases. Therefore, for example, a large-scale distributed database system can be configured with a relatively low performance computer.

  The above and other objects, features, aspects and advantages of the present invention will become apparent from the following detailed description of the present invention taken in conjunction with the accompanying drawings.

1 is a diagram illustrating a configuration of a control system 1 according to Embodiment 1. FIG. 3 is a diagram illustrating a hardware configuration of the first computer 100 according to the first embodiment. FIG. FIG. 3 is a block diagram illustrating functions of the first computer 100 in the first embodiment. 4 is a flowchart showing a process in which the data update management unit 15 in Embodiment 1 calculates an update sequence number value by the update value calculation unit 18 and updates a data record. FIG. 10 is a diagram illustrating a process in which replication is performed between computers including a database in the first embodiment, and an update sequence number value is reflected in another database. 6 is a diagram illustrating an example in which the contents of a data record are made identical to a database of another computer by the replication processing in the first embodiment. FIG. 6 is a diagram showing a replication process when a plurality of computers in the first embodiment updates a data record. FIG. 12 is a flowchart illustrating processing in which each computer in Embodiment 2 changes a priority value. It is a figure which shows the specific example of the process in which the computer in Embodiment 2 changes a priority value. FIG. 10 is a second flowchart illustrating processing in which each computer in Embodiment 2 changes a priority value.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, the same parts are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated.

<Embodiment 1>
The computer of this embodiment includes a database, and a plurality of computers are installed to constitute a database system. The database system performs replication for making the contents of the database of each computer identical by the multi-master method. The database system is used in the control system 1, for example. The control system 1 performs various controls based on the data of this database system.

  FIG. 1 is a diagram showing a configuration of a control system 1 in the first embodiment. Referring to FIG. 1, the control system 1 includes a plurality of bases (first base 5A, second base 5B, third base 5C, fourth base 5D) and a plurality of computers (first computer 100, second computer). 200, the third computer 300), and the plurality of bases and the plurality of computers are connected to each other by the network 2. Each computer such as the first computer 100 includes a database, and a database system is configured by these computers. The bases such as the first base 5A perform monitoring of monitoring targets and commands to each device using data stored in the database system.

  The control system 1 is, for example, a monitoring control system, and a large number of monitoring devices or the like sequentially acquire data, and these acquired data are stored in a database system (first computer 100, second computer 200, third computer 300). And the operation base (first base 5A etc.) performs monitoring based on the data of the database system and gives various commands.

  The monitoring control system is, for example, a system control system for stably supplying electric power generated in a power plant to consumers such as factories. The system control system monitors the status of the system and collects monitoring values such as the maximum and minimum values of voltage and current, changes in phase, and ON / OFF status of the circuit breaker (Circuit Breaker). Update data records in the database system with the updated data. Since the system control system requires stable continuous operation, a computer constituting the database system and a network for connecting the computers are multiplexed. For example, by installing each computer and operation base (central power supply command station, grid power supply command station, etc.) geographically dispersed, disasters occur and some computers and operation bases can function. Even if it runs out, it is possible to continue operation by switching computers and operation bases.

<Configuration>
A hardware configuration and a functional configuration of a computer (first computer 100 or the like) according to the present embodiment will be described. FIG. 2 is a diagram illustrating a hardware configuration of the first computer 100 according to the first embodiment. Referring to FIG. 2, the first computer 100 includes a processor 61, a memory 62, an HDD 63, a communication interface 64, and the like.

  The processor 61 reads the program, controls the operation of the first computer 100 according to the read program, and causes the first computer 100 to function as a database management system. The memory 62 is a volatile storage area for holding data processed in the first computer 100. The HDD 63 is a non-volatile storage area for the first computer 100 to function as a database management system, and holds a plurality of data records. The communication interface 64 functions as an interface for the first computer 100 to communicate with an external device.

  FIG. 3 is a block diagram illustrating functions of the first computer 100 according to the first embodiment. A plurality of computers (the first computer 100, the second computer 200, and the third computer 300) are connected by the network 2 to constitute a database system. Since the functional configuration of each computer is the same, the configuration of the first computer 100 will be described in detail. Referring to FIG. 3, the first computer 100 includes a database file 11, an update sequence number management file 12, a priority management file 13, a priority management unit 14, a data update management unit 15, and a data transmission / reception unit 16. And an update request source program 17.

  The database file 11 is a storage area for storing a plurality of data records constituting the database.

  The update sequence number management file 12 is a storage area for storing an update sequence number value indicating the update state of the data record stored in the database file 11. The update sequence number value indicates the update state of the data record by a serial number. The update sequence number management file 12 stores an update sequence number value for each data record stored in the database file 11.

  The priority management file 13 is a storage area for storing a priority value indicating the priority assigned to the first computer 100. The priority value is used to determine which database content is to be prioritized and reflected in each database when data updates conflict in a plurality of databases. In a database system consisting of a plurality of computers, priority values are set in association with each database of each computer (or each specific group of each database), and each database is assigned a unique priority value. ing.

  The range of values that the priority value can take is, for example, an integer value with the total number of databases constituting the database system as a maximum value. For example, when there are three computers, the maximum value that the priority value can take is 3, and each computer is assigned one of the values from 1 to 3 as the serial number and stores it as a priority value. In this embodiment, the computer can also operate in an invalid state that does not permit database update. The invalid state is, for example, an initial startup state of the computer or a state in which a priority value described later is being changed. During this invalid state, the computer operates with the priority value set to 0.

  The priority management unit 14 changes the priority value assigned to the database of each computer through communication with other computers.

  The update request source program 17 is a program for instructing the data update management unit 15 to perform database update processing in the first computer 100, and issues a database update request to the data update management unit 15.

  The data update management unit 15 includes an update value calculation unit 18 and a record update control unit 19. The data update management unit 15 writes data to the database file 11, calculates and updates an update sequence number corresponding to the data record, and other computers. Control replication with other databases.

  When the update value calculation unit 18 receives a data write request from the update request source program 17 and writes data to the database file 11, the update value calculation unit 18 calculates an update sequence number value after the data is written to the database file 11 and calculates the calculated update. The serial number value is stored in the updated serial number management file 12. The update value calculation unit 18 calculates the update sequence number value associated with the data writing based on the update sequence number value of the data record related to the data before update and the priority value stored in the priority management file 13.

  The record update control unit 19 receives a data write request from the update request source program 17 and performs a process of writing data to the database file 11. In addition, the record update control unit 19 controls a replication process for matching the contents with the database of another computer. When the data update control unit 19 receives the data record and the update sequence number value of the data record from another computer by the data transmitting / receiving unit 16, the update sequence number value stored in the update sequence number management file 12 for the received data record, The received update sequence number value is compared, and whether or not the database file 11 is updated using the received data record and the update sequence number value is controlled based on the size of the update sequence number value.

  Each component of the second computer 200 and the third computer 300 has the same configuration as the first computer 100 except for reference numerals. Computers such as the first computer 100, the second computer 200, and the third computer 300 are connected to each other by the network 2, and the contents of the database are made identical by transmitting and receiving data records and update sequence numbers.

<Operation>
Regarding the operation of the first computer 100, a process in which the update value calculation unit 18 of the first computer 100 calculates the update sequence number value and a process in which the first computer 100 replicates the contents of the database with other computers are specifically described. explain.

(Process to calculate the update sequence number when updating data)
FIG. 4 is a flowchart illustrating processing in which the data update management unit 15 according to the first embodiment calculates the update sequence number value by the update value calculation unit 18 and updates the data record. The data update management unit 15 receives a data write request from the update request source program 17, calculates the update sequence number value by the update value calculation unit 18, and records the data record of the database file 11 by the record update control unit 19. Start the update process.

  In step S <b> 21, the update value calculation unit 18 receives data to be updated in response to a data write request from the update request source program 17.

  In step S <b> 22, the update value calculation unit 18 extracts the priority value set for the first computer 100 from the priority management file 13.

  In step S23, the update value calculation unit 18 determines whether the priority value extracted from the priority management file 13 is valid (priority value) in order to determine whether the update of the data record is permitted or invalid. Is not 0) or is invalid (whether the priority value is 0). If the priority value is valid, the update value calculation unit 18 performs step S24. If the priority value is invalid, the update value calculation unit 18 proceeds to step S26, and the first computer 100 updates the update request source program. In response to the write request from 17, the data update management unit 15 does not update the data record, and the process ends.

  In step S24, the update value calculation unit 18 reads the update sequence number value (current update sequence number value) of the data record before update from the update sequence number management file 12, and the read update sequence number value and the priority read in step S22. Based on the value, the update sequence number value after the update of the data record is calculated according to the number of updates of the data record according to the following equation (1).

Update sequence number after update = {INT (current update sequence number / total number of databases constituting the database system) + 1} x total number of databases constituting the database system + (priority value -1) (1)
In the formula (1), “INT (X)” indicates a maximum integer value not exceeding X. Referring to Equation (1), in the first embodiment, the update value calculation unit 18 of each computer uses the product of the number of data record updates and the number of databases (computers) constituting the database system as a reference value. The updated serial number value after the update of the data record is calculated by correcting the value based on the priority value assigned to each computer with respect to the reference value.

  In the first embodiment, the update value calculation unit 18 calculates the update sequence number value after updating the data record by adding “number obtained by subtracting 1 to the priority value” to the reference value. The update value calculation unit 18 calculates INT (current update sequence number / total number of databases constituting the database system) in Expression (1) to obtain the number of updates before updating the data record. Since a unique value is assigned to each computer as the priority value, the update sequence number value calculated by each computer is a value unique to each computer. Therefore, by referring to the update sequence number value associated with the data record, it is possible to determine which computer has updated the data record.

  In step S <b> 25, the record update control unit 19 updates the data record of the database file 11 based on the data requested to be written from the update request source program 17. The update value calculation unit 18 stores the update sequence number value calculated in step S24 in the update sequence number management file 12. The record update control unit 19 transmits the updated data record and the updated update sequence number value to other computers (second computer 200, third computer 300, etc.) by the data transmitting / receiving unit 16.

In step S26, the data record update process is terminated.
(Replication processing when data is received from other computers)
When the first computer 100 receives the data updated by the other computer by the data transmission / reception unit 16 and the update sequence number of the data, the first computer 100 uses the received data as the first computer in order to make the contents of the database identical to the other computer. The data update management unit 15 controls whether or not to be reflected in the 100 database files 11.

  The data update management unit 15 reads the update sequence number value stored in the update sequence number management file 12 for the data record at the position where the received data is stored in the database file 11, and the read update sequence number value and the received update sequence number value And compare. When the received update sequence number value is larger than the update sequence number value read from the update sequence number management file 12, the data update management unit 15 updates the database file 11 with the received data and stores it in the update sequence number management file 12. The updated serial number value is updated.

  In addition, when the update sequence number value of the data received from another computer is smaller than the update sequence number value read from the update sequence management file 12, the data update management unit 15 stores the data stored in the database file 11 in the own computer. Prioritize the record value. Therefore, the data update management unit 15 discards the received data and the update sequence number value, and transmits the data record stored in the database file 11 and the update sequence value corresponding to the data record to another computer via the network 2. To do. As a result, the first computer 100 performs a replication process with a database of another computer constituting the database system.

(Specific examples of replication processing)
A replication process performed between a plurality of computers as described above will be described using a specific example.

  FIG. 5 is a diagram illustrating processing in which replication is performed between computers including a database and the update sequence number value is reflected in another database in the present embodiment in the first embodiment. FIG. 5 shows a case where the total number of computers having a database is 3, and shows an example of specific changes in the update sequence number management file 12, the update sequence number management file 22, and the priority management file 33 of each computer. In this example, the first computer 100 holds the priority value “1” in the priority management file 13, and the second computer 200 holds the priority value “2” in the priority management file 23, The third computer 300 holds the priority value “3” in the priority management file 33.

  In step S31, all the update sequence numbers managed in the update sequence number management file 12, the update sequence number management file 22, and the update sequence number management file 32 for the same record are in the initial state (the value of the update sequence number value is 0). Suppose there is. Here, it is assumed that the first computer 100 has made a data update request by the update request source program 17.

  In step S32, the first computer 100 updates the record of the database file 11 by the data update management unit 15 and performs the process of updating the update sequence number value (step S24 and step S25 in FIG. 4), and the update sequence number management file 12 The update sequence number value “3” is stored in.

  In step S <b> 33, the first computer 100 performs the same processing (replication processing) of the stored contents of the database as the other computers by the processing of the record update control unit 19. As a result of this replication processing, the update sequence number value “3” held in the update sequence number management file 12 for the record updated by the first computer 100 is updated in the update sequence number management file 22 of the second computer 200 and the third computer 300. It is reflected in the serial number management file 32.

  In step S34, it is assumed that the second computer 200 has updated the record. The second computer 200 performs a process for updating the update sequence number value (steps S24 and S25 in FIG. 4), calculates the update sequence number “7”, and stores the update sequence number “7” in the update sequence number management file 22. To do.

  In step S <b> 35, the second computer 200 performs a replication process with another computer by the process of the record update control unit 29, and the update sequence number value “stored in the update sequence number management file 22 for the record updated by the second computer 200“. 7 ”is reflected in the update sequence number management file 12 of the first computer 100 and the update sequence number management file 32 of the third computer 300.

  In step S36, the third computer 300 updates the record. The third computer 300 performs a process for updating the update sequence number value (steps S24 and S25 in FIG. 4), calculates the update sequence number “11”, and stores the update sequence number “11” in the update sequence number management file 32. To do.

  In step S <b> 37, the third computer 300 performs replication processing with another computer by the processing of the record update control unit 39, and the update sequence number value “held in the update sequence number management file 32 for the record updated by the third computer 300 is displayed. 11 ”is reflected in the update sequence number management file 12 of the first computer 100 and the update sequence number management file 22 of the second computer 200.

  As described above, the first computer 100, the second computer 200, and the third computer 300 calculate the update sequence number value based on the update count of the data record and the priority value uniquely assigned to each computer. ing. As a result, the update sequence number value is a value unique to the computer that updated the data record. For example, in the example of FIG. 5 described above, a priority value is assigned to each computer by a serial number, and a reference value determined according to the number of data record updates is corrected with the priority value. For example, when the first computer 100 updates the data record, the update sequence number value is “3 × n” (n is an arbitrary positive integer), and when the second computer 200 updates the data record, the update sequence number value is When it becomes “3 × n + 1” and the third computer 300 updates the data record, the update sequence number value becomes “3 × n + 2”. By referring to the priority value, each computer can determine which computer has updated the data record.

  In addition, by calculating the reference value based on the product of the number of updates of the data record and the number of computers in this way, by calculating the updated serial number value by correcting the calculated reference value by the priority value of each computer, The update sequence number value can be made unique to each computer by simple processing.

  FIG. 6 is a diagram illustrating an example in which the content of the data record is made identical to the database of another computer by the replication processing in the first embodiment. FIG. 6 shows an example in which the contents of the data records of the first computer 100 and the second computer 200 are made identical based on the update sequence number value. Replication processing is performed when each computer updates a data record and sends the updated data record and update sequence number value to each other computer, or when each computer receives a data record and update sequence number value from another computer. This is performed by each computer when the communication state is recovered after an abnormality occurs in the network 2 and communication with another computer becomes impossible.

  In step S41, the first computer 100 compares the update sequence number management file (update sequence number management file 22) of another computer (second computer 200) with the update sequence number management file 12 of its own computer for each data record. As a result of the comparison, if the value of the update sequence number value is different, the first computer 100 gives priority to the data record having a large update sequence number value. For example, in the example of FIG. 6, the update sequence number values of the record 1 and the record 3 are different between the first computer 100 and the second computer 200. The first computer 100 refers to the update sequence number value of each data record, and determines that the database file 11 is prioritized for the record 1 and the database file 21 is prioritized for the record 3.

  In step S42, replication processing is performed, and between the first computer 100 and the second computer 200, the storage content of the database file 11 is given priority over the database file 21 for the record 1, and the database file 21 for the record 3 is recorded. Is prioritized over the database file 11. As each computer updates the data record and the update sequence number, the contents of the database of each computer are made identical.

  As described above, in this embodiment, the replication process can be performed by a simple process of comparing the update sequence number values stored in the update sequence number management file (update sequence number management file 12 or the like) by each computer. Operation can be made more efficient in a large database.

(When data records are updated on each computer)
When each computer updates the data record, the data record of the computer with the highest priority value is given priority. FIG. 7 is a diagram showing a replication process when a plurality of computers in the first embodiment updates a data record. For example, when multiple computers update data records almost simultaneously, or when a communication connection between computers fails and each database updates data records during periods when replication processing is difficult . In the example of FIG. 7, an example is shown in which the second computer 200 and the third computer 300 update the same data record.

  In step S51, the contents of the database of each computer are made identical, and the update sequence value “11” is held in the update sequence number management file of each computer. Thereafter, it is assumed that the second computer 200 and the third computer 300 each update the same data record.

  In step S52, as a result of the second computer 200 updating the database file 21 and the third computer 300 updating the database file 31, the update sequence number management file 22 of the second computer 200 holds the updated sequence number “13”. The update sequence number management file 32 of the third computer 300 holds the update sequence number value “14”.

  In step S53, replication processing is performed, and the contents of the data record of the third computer 300 having the largest update sequence number value are matched with those of other computers.

  As described above, according to the present embodiment, even if each computer updates a data record almost simultaneously, the replication process can be performed by a simple process of comparing the update sequence number values.

<Embodiment 2>
In the first embodiment, the priority value assigned to each computer has been described as being fixed. However, each computer may change the value of the priority value. For example, the priority value of each computer may be changed when a new computer is added or when the performance of the computer is improved. The processing for changing the priority value is performed by the priority management unit (priority management unit 14, priority management unit 24, priority management unit 34, etc.) of each computer. Since each computer is assigned a unique priority value, it is necessary for each computer to change the priority value so that a plurality of computers do not have the same priority value. Specifically, when each computer changes the priority value, the priority management unit of each computer exchanges the value of the priority value with each other, and the value of the priority value to be changed by each computer is already used. If it is, the request to change the priority value is ignored. Further, the priority value is changed after setting the value of the priority value of any computer to an invalid state (setting the value of the priority value to 0) so that the computers do not have the same priority value.

  FIG. 8 is a flowchart illustrating processing in which each computer in the second embodiment changes the priority value. In the processing shown below, in order from the computer with the highest priority value, it is determined whether or not the own computer is the target for changing the priority value, and the priority value of each computer is re-established while decreasing the priority value to be changed. Set.

  In step S101, the computer communicates with other computers through the priority management unit (priority management unit 14 or the like), and acquires a reset value of the priority value that each computer should change. For example, a priority is assigned to each computer, and the priority value is reset so that the priority value of a computer with a higher priority becomes larger. The priority of each computer is determined based on, for example, each computer's MAC (Media Access Control) address, IP (Internet Protocol) address, each computer's serial number, each computer's performance value, and the importance of each computer.

  In step S103, each computer sets a maximum priority value (total number of computers) in a flag counter (change target device information) indicating a device to be changed.

  In step S105, each computer compares the value of the priority value managed by the priority management unit of its own computer with the change target device information, and whether or not the priority value matches the change target device information. Judging. If the determination result in step S105 is affirmative (if the priority value matches the change target device information), the computer executes the process of step S107. In the case of a negative determination result in step S105 (the value of the priority value does not match the change target device information), the computer executes the process of step S113.

  In step S107, the computer determines whether it is necessary to change the value of the priority value of its own computer based on the reset value of the priority value of each computer acquired in step S101. When the setting of the priority value of the own computer is not changed, the computer executes the process of step S109. When the setting of the priority value of the own computer is changed, the computer executes the process of step S111.

  In step S109, the computer does not change the value of the priority value and notifies the other computers that the priority value change processing in the own computer has been completed.

  In step S111, the computer sets the priority value of its own computer to an invalid state (priority value = 0), and notifies other computers that the value of the priority value has been invalidated.

  In step S <b> 113, the computer waits until there is a priority value notification from another computer, and receives this notification.

In step S115, each computer decrements the value of the change target device information.
In step S117, each computer compares the priority value managed by the priority management unit of its own computer with the change target device information, and whether or not the priority value matches the change target device information. Is determined to determine whether the own computer is a computer whose priority value is to be changed. In the case of a negative determination result in step S117 (when the priority value does not match the change target device information), the computer executes the process of step S119. In the case of a positive determination result in step S117 (when the priority value matches the change target device information), the computer executes the process of step S121.

  In step S119, the computer waits until there is a notification regarding the priority value from another computer, and receives this notification.

  In step S121, the computer changes the value of the priority value of the own computer according to the reset value of the priority value, and notifies the other computers that the priority value has been changed.

In step S123, each computer decrements the value of the change target device information.
In step S125, each computer refers to the reset value of the priority value and the value of the priority value changed in each computer, and determines whether or not the priority values of all the computers have been changed. If the priority values of all the computers have been changed (YES in step S125), each computer ends the priority value change process. When the change of the priority value of all the computers has not been completed (NO in step S125), each computer performs the process of step S117.

  FIG. 9 is a diagram illustrating a specific example of processing in which each computer in the second embodiment changes the priority value. In the example of FIG. 9, an operation example in the case where three computers (first computer 100, second computer 200, and third computer 300) change the priority value is shown.

  In step S61, the priority management unit of each computer holds the priority value “1” in the priority management file 13 of the first computer 100, and the priority value “2” in the priority management file 23 of the second computer 200. And the priority value “3” is held in the priority management file 33 of the third computer 300. By resetting the priority value, the priority value assigned to the first computer 100 is “2”, the priority value assigned to the second computer 200 is “3”, and the priority value assigned to the third computer 300 is “3”. It shall be changed to “1”.

  In step S62, the priority management unit 34 of the third computer 300 performs the processing of steps S105 and S107, and sets the priority value to an invalid state (priority value = 0).

  In step S63, the priority management unit 34 of the third computer 300 performs the process of step S111, and notifies the other computers that the priority value is in an invalid state (priority value = 0). The first computer 100 and the second computer 200 perform the process of step S113 and receive a notification that the third computer 300 has made the priority value invalid.

  In step S64, the second computer 200, when receiving the notification from the third computer 300 in step S63, if there is data to be updated in the data record with the data received from the third computer 300, the database file The data is reflected in 21.

  In step S <b> 65, the second computer 200 performs the processes of steps S <b> 117 and S <b> 121 to cause the priority management file 23 to retain the priority value “3”.

  In step S66, the second computer 200 performs the process of step S121 to notify another computer that the second computer 200 has changed the value of the priority value. The first computer 100 and the third computer 300 perform the process of step S119 and receive a notification that the second computer 200 has changed the priority value.

  In step S67, the first computer 100 determines that the data received from the second computer 200 before the notification from the second computer 200 in step S66 has data to be updated in the data record. 11 reflects the data.

  In step S <b> 68, the first computer 100 performs the processing of steps S <b> 117 and S <b> 121 to hold the priority value “2” in the priority management file 13.

  In step S69, the first computer 100 performs the process of step S121 to notify another computer that the first computer 100 has changed the priority value. The second computer 200 and the third computer 300 perform the process of step S119 and receive a notification that the first computer 100 has changed the priority value.

  In step S70, the third computer 300 determines that the data received from the first computer 100 before the notification from the first computer 100 in step S69 has data to be updated in the data record. The data is reflected in 31.

  In step S <b> 71, the third computer 300 performs the processing of steps S <b> 117 and S <b> 121 to hold the priority value “1” in the priority management file 33.

  With the above processing, each computer can change the priority value without duplicating the priority value.

  FIG. 10 is a second flowchart illustrating a process in which each computer in the second embodiment changes the priority value. Even if the processing shown in FIG. 8 and FIG. 9 is not used, each computer prohibits / permits the update of its own data before and after the setting of the new priority value to each computer. The procedure for avoiding duplicate values is shown. Each computer instructed with the new priority value in step S201 executes the flow of FIG. In step S202, the computer prohibits updating of self data. As a result, after the update data of other computer data that has not yet been reflected in step S203 is reflected, new update data is not reflected, so the update sequence number value by changing the priority value of each computer is changed. Duplication can be avoided. In step S204, the self priority value is changed to a designated value. Next, in step S205, the own priority value change completion is transmitted. Each computer waits until it can be confirmed in step S206 that another computer has completed the change of the priority value, and after the change of the priority value of all the computers is completed, the update of the self data is permitted in step S207, and step S209 is performed. End with.

  As explained in each embodiment, in a database system that applies a multi-master method that can be updated from any distributed database, time information of computers that are easily affected by time lag, reference to complex parameters, etc. Is not required, and contention at the time of updating the data record can be avoided by simply performing the process of comparing the update sequence number values. Therefore, even in an environment where time synchronization is not possible or a database system is configured with a low-performance computer, it is possible to configure a large-scale distributed database system using the multi-master method.

  In the above description, an example of a database system including three computers has been mainly described, but the number of databases is not limited to three. Moreover, the target managed by the priority value is not limited to the entire database, and may be a group of a specific data record.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiment. However, the present invention is not limited to the embodiment, and various modifications can be made without departing from the scope of the invention. Needless to say. The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

  DESCRIPTION OF SYMBOLS 1 Control system, 2 Network, 5A 1st base, 5B 2nd base, 5C 3rd base, 5D 4th base, 11 Database file, 12 Update serial number management file, 13 Priority management file, 14 Priority management part, 15 Data update management unit, 16 data transmission / reception unit, 17 update request source program, 18 update value calculation unit, 19 record update control unit, 21 database file, 22 update sequence number management file, 23 priority management file, 24 priority management unit, 25 data update management unit, 26 data transmission / reception unit, 27 update request source program, 28 update value calculation unit, 29 record update control unit, 31 database file, 32 update sequence number management file, 33 priority management file, 34 priority management unit , 35 Data update management unit, 36 Data transmission / reception Parts, 37 update request source program, 38 update value calculation unit, 39 record update control unit, 61 Processor, 62 memory, 63 HDD, 64 a communication interface, 100 a first computer, 200 the second computer, 300 third computer.

Claims (6)

A computer that constitutes a database system that performs replication among multiple computers by a multi-master method,
A data record holding unit for holding a plurality of records;
A storage unit for storing an update value indicating an update status of the record for each record;
When updating a record in its own computer, an update value specific to the own computer is calculated according to the number of updates of the record to be updated, based on the update value of the record before the update and the priority value specific to the own computer. A calculating unit to
A record updated in the own computer and an update value unique to the calculated own computer for the record are transmitted to another computer constituting the database system, and the record updated in the other computer from the other computer And a transmission / reception processing unit that receives an update value of the record,
When the record and the update value of the record are received from the other computer, the plurality of computers are compared by comparing the received update value with the update value stored in the storage unit for the received record. And a controller that performs replication between
calculator. The process of calculating the update value specific to the computer by the calculation unit is as follows:
Obtain the number of updates of the record based on the update value of the record before the update, calculate a reference value based on the product of the obtained number of updates and the number of computers constituting the database system, and calculate The computer according to claim 1, wherein the calculated reference value is corrected by the priority value. The storage unit
The network host computer in a computer group constituting the database system by the connection is configured to store a unique priority value,
The computer according to claim 2. The priority value is a serial number assigned to each computer constituting the database system,
The calculation unit calculates an update value specific to the own computer by adding or subtracting the priority value to the calculated reference value.
The computer according to claim 3. The calculator is
Accepts request of resetting the priority value stored in the Yu Sakichi storage unit, the communications with other computers constituting the database system further comprises a priority value manager to change the priority value among the computer,
The computer according to any one of claims 1 to 4. The priority value management unit
Based on the device-specific information of each computer constituting the database system, a priority value to be reset in the own computer is acquired.
The computer according to claim 5.

Images