JP2001034518A - Master-slave relation information synchronization system in decentralized database - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make efficiently synchronizable updated data in a decentralized database system of multiple servers having a master-slave relation between a master table and a slave table while maintaining the master-slave relation. SOLUTION: A main table (dummy) 93 is provided in a relational database (slave) 9 on a slave server 7 and the master-slave relation is defined between the master table (dummy) 93 and a slave table (copy) 92. A log monitor means 5 periodically monitors whether or not an update log file 4 is updates according to update log information outputted when a master table (master) 31 is updated in a relational database (master) on a master server 1. An update reflecting process means 6 reflects the update contents on the master table (dummy) 93 in the relational database (slave) 9 in the slave server 7 according to the latest update log information of the update log file 4 when the log monitor means 5 detects the update of the update log file 4.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a master-slave relationship information synchronization method in a distributed database system, and more particularly to a master table (also referred to as a reference table or parent table) and a slave table (referring to some data items of the master table). The present invention relates to a master-slave relationship information synchronization method in a distributed database system including a plurality of servers having a master-slave relationship with a related table (also referred to as a dependent table or a child table). Here, the master-detail relationship of the table means that the dependent table is included in the definition of the referential integrity constraint that refers to the primary key (refers to a data item that is unique within the table) of the primary table (refers to a data item that is unique in the table). (Referred to as a column or a set of columns) is referred to as a relationship between records in the main table and records in the sub-table.

[0002]

2. Description of the Related Art In a conventional distributed database system, for example, a master table (master) and a slave table (master) are defined, and a master-slave relationship is defined between the master table (master) and the slave table (master). A master server having a relational database (master) and an RDBMS (master) managing the relational database (master);
Relational database (slave) in which main table (copy) and slave table (copy) are defined, and RDB managing this relational database (slave)
A slave server having an MS (slave) is connected to a LAN
(Local Area Network) or the like.

In such a conventional distributed database system, the slave server cannot define the master-slave relationship between the main table (copy) and the slave table (copy) (the primary key cannot be defined in the master table (copy)). Therefore, the data consistency of the data must be guaranteed on the application program side, for example, by having information on the master-slave relationship in the data record. That is, since the master-slave relationship was held as information in the record, the application program had to perform processing to maintain data consistency.
A different update process had to be created for the master server and the slave server, taking into account the differences in the environment.

[0004]

A first problem is that it is difficult to efficiently synchronize updated data while maintaining a master-slave relationship. The reason is that it is necessary for the application program to perform a process for associating a master-slave relationship separately from the process for synchronizing the contents of the master table and the slave table.

A second problem is that when a new subordinate table is added to the main table, the subordinate table is defined (master server,
This means that it is necessary to update data (slave server) together with the implementation of consistency assurance on the application program side (server only). The reason for this is that the application program must perform processing for maintaining data consistency.

[0006] By the way, tables in a relational database can be related by a common column. Therefore, rules for managing column relations must be observed. However, as a function for guaranteeing that these relations are maintained, relational tables are used. Database management system (hereinafter RD)
(Abbreviated as BMS). By using the referential integrity constraint, a master table (dummy) is defined in the slave server, and a master-slave relationship is defined between the master table (dummy) and the slave table (copy), thereby enabling data to be defined. It is conceivable to ensure consistency.

However, the master table (dummy) is defined in the slave server, and the master table (dummy) and the slave table (copy) are defined.
If a master-servant relationship is defined between the master table and the master table (dummy), the data consistency between the main table (master) and the main table (dummy) should be further verified. You have to consider what you want to do.

An object of the present invention is to maintain a master-slave relationship for updated data in a distributed database system comprising a plurality of servers having a master-slave relationship between a master table and a slave table by using referential integrity constraints. It is an object of the present invention to provide a master-slave relationship information synchronization method in a distributed database system capable of achieving efficient synchronization.

As a prior art document, for example, there is JP-A-5-135113. In the duplicate data management system disclosed in this publication, a client is provided with a derivation destination management table and a derivation source management table, In response to the database operation language request from, the management table control means searches the derivation destination management table and the derivation source management table, and checks whether or not the data to be accessed is managed as duplicate data. Management control was performed. Therefore, this duplicated data management system is completely different from the present invention on the premise that the same data is retained in the relational database (master) and the relational database (slave) in duplicate, and the purpose, configuration and effect are completely different. It is.

[0010]

According to the master-slave relationship information synchronizing method in the distributed database system of the present invention, a master table and a slave table are defined, and the master table and the slave table are defined. Relational database (master) in which master-slave relationship is defined between RDB and RDB managing the relational database (master)
A master server having an MS (master) and a slave server having a relational database (slave) in which a master table (copy) and a slave table (copy) are defined and an RDBMS (slave) managing the relational database (slave) In a distributed database system connected via a line, a master table (dummy) is further defined and a master-slave relationship is defined between the master table (dummy) and the slave table (copy). A database (slave), an update log file provided in the master server, and storing update log information when the main table (master) of the relational database (master) is updated by the RDBMS (master); Provided in the master server,
A log monitoring unit that periodically monitors whether the update log file has been updated; and a unit that, when the update of the update log file is detected by the log monitoring unit, updates the log information of the update log file. And the main table (dummy) in the relational database (slave)
And an update reflection processing unit for reflecting the update content to the update.

On the other hand, according to the recording medium of the present invention, a log monitoring means for periodically monitoring whether or not the update log file has been updated, and the update of the update log file is detected by the log monitoring means. A program for causing the main table (dummy) in the relational database (slave) to reflect update contents based on the latest update log information of the update log file, and to function as update reflection processing means. Record

DETAILED DESCRIPTION OF THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of a distributed database system to which a master-slave relationship information synchronization system according to a first embodiment of the present invention is applied. This distributed database system includes a master server 1 as a master server machine, a slave server 7 as a slave server machine, and a client 10 as a client machine.
Are connected via a LAN.

The master server 1 is an RDBMS (master)
2, a relational database (master) 3, an update log file 4, a log monitoring unit 5, and an update reflection processing unit 6.

[0015] Relational database (master) 3
Has a master table (master) 31 and a slave table (master) 32 storing data. A master-slave relationship between tables is defined between the master table (master) 31 and the slave table (master) 32.

Here, the main table (master) 31 and the sub table (master) 32 are separate tables because the tables are normalized. The normalization is a method of eliminating data redundancy and making the data structure consistent and consistent and easy to understand. When the normalization is performed, only the primary key portion is duplicated, but the duplication of the other data items can be omitted, and the processing at the time of updating the data is simplified. Note that the primary key and the foreign key are both set in the columns of the table when the table is normalized.

The RDBMS (master) 2 checks the validity of the update process (SQL) and checks whether the update process (SQL) is correct when the update process to the main table (master) 31 and the slave table (master) 32 is requested. This is a relational database management system that checks whether or not the referential integrity constraint is violated, and notifies the RDBMS (slave) 8 that it has been updated. Also, R
When the DBMS (master) 2 is notified from the RDBMS (slave) 8 that the main table (copy) 91 and the sub table (copy) 92 have been updated, the main table (master) 31 and the sub table (Master) 32 performs equivalent update processing.

The update log file 4 is a file for storing update log information when the main table (master) 31 is updated in the relational database (master) 3. The update log file 4 stores the update content of the main table (master) 31 as update log information.

The log monitoring means 5 updates the update log file 4 with the update log information output from the RDBMS (master) 2 when the main table (master) 31 in the relational database (master) 3 is updated. Is regularly monitored to determine if it has been updated.
This is an application program unit that calls the update reflection processing unit 6.

The update reflection processing means 6, when called from the log monitoring means 5, uses the main table (slave) 9 in the relational database (slave) 9 of the slave server 7 based on the latest update log information of the update log file 4. (Dummy) 93 is an application program means for reflecting the updated contents.

The slave server 7 includes an RDBMS (slave) 8 and a relational database (slave) 9
And

Relational database (slave)
Reference numeral 9 denotes a main table (copy) 91 which is a copy of the main table (master) 31, a sub table (copy) 92 which is a copy of the sub table (master) 32, and data having the same contents as the main table (master) 31. And a main table (dummy) 93 having.

The main table (dummy) 93 is a main table (copy) 9
1 is a table defined by a name different from 1.

Main table (copy) 91 and slave table (copy)
Reference numeral 92 denotes a main table (master) 3 defined in the relational database (master) 3 in the RDBMS (master) 2
1 and the master 32, and have the same data as the RDBMS (master) 2 and RDB
It is updated synchronously by the referential integrity constraint of the MS (slave) 8.

However, a master-slave relationship is not defined between the master table (copy) 91 and the slave table (copy) 92. Although the contents match with the master table (master) 31 and the slave table (master) 32, a master-slave relationship is not defined between the master table (copy) 91 and the slave table (copy) 92. In a distributed database system, since a primary key cannot be set for a main table (copy) 91, a master-slave relationship between a main table (copy) 91 and a dependent table (copy) 92 can be defined. Because you can't.

For this reason, in the present embodiment, a master-slave relationship is defined between the master table (dummy) 93 and the slave table (copy) 92. That is, in order to define the master-slave relationship between the master table (copy) 91 and the slave table (copy) 92 in the relational database (slave) 9, a table having data having the same contents as the master table (copy) 91 is called an alias. Is defined as a master table (dummy) 93 to define a master-slave relationship with a slave table (copy) 92. Thus, the main table (dummy) 93
Is created and a master-slave relationship is defined between the master and the slave table (copy) 92 in order to use the master-slave relationship to realize the following three points.

1) As means for maintaining data consistency, R
DBMS (master) 2 and RDBMS (slave) 8
To take advantage of the referential integrity constraints of

2) Relational database (master) 3 and relational database (slave)
In the process of updating data for the main table (master) 31 and the subordinate table (master) 32 and the main table (copy) 91 and the subordinate table (copy) 92 in 9, the user is aware of the difference in the environment such as the master server 1 and the slave server 7. Not to be.

3) Although the same as 1) and 2),
Even when a new subordinate table (master) 32 and a new subordinate table (copy) 92 to the main table (master) 31 and the main table (copy) 91 are newly added, only the definition of the subordinate table is used, and data consistency is maintained. RDBMS (master) 2 and RDBMS
(Slave) In order to use the referential integrity constraint of 8 and not to be aware of differences in the environment of the data update processing.

The RDBMS (slave) 8 checks the validity of the update process (SQL) and checks whether the update process (SQL) is correct when the update process to the main table (copy) 91 and the slave table (copy) 92 is requested. This is a relational database management system that checks whether or not a reference integrity constraint has been violated, and notifies the RDBMS (master) 2 that it has been updated. Also, RDB
When the MS (slave) 8 is notified from the RDBMS (master) 2 that the main table (master) 31 and the slave table (master) 32 have been updated, the main table (copy) is used.
The same update processing is performed on the data 91 and the dependent table (copy) 92.

The client machine 10 has an update processing means 11.

The update processing means 11 updates the main table (master) 31 and the slave table (master) 32 in the relational database (master) 3 of the master server 1 via the LAN, or updates the relational database (slave) of the slave server 7. ) 9 is an application program for updating the main table (copy) 91 and the slave table (copy) 92 in 9. The update process here is
This indicates insertion of data into the table (Insert), update of data in the table (Update), and deletion of data in the table (Delete).

Referring to FIG. 2, the processing of the master-slave relationship information synchronization method in the distributed database system according to the first embodiment includes an update log information output step A1, an update log file update determination step A2, and an update reflection processing. It comprises a means calling step A3 and a main table (dummy) updating step A4.

FIG. 3 shows a relational database (master) 3, a relational database (slave) 9
FIG. 4 is a diagram illustrating the contents of an update log file 4;

Next, the operation of the master-slave relationship information synchronization method in the distributed database system according to the first embodiment will be described with reference to FIGS.

As shown in FIG. 3, for example, the main table (master) 31 and the sub table (master) 32 have "code 000".
RDBMS (master) when the data of "2" is added
2, the update log information of the main table (master) 31 is written to the update log file 4 (step A1).

The log monitoring means 5 periodically monitors whether or not the update log file 4 has been updated, and upon detecting that the update log file 4 has been updated (step A2), updates are reflected. The processing means 6 is called (step A3).

The update reflection processing means 6 adds data of "code 0002" to the main table (dummy) 93 based on the latest update log information of the update log file 4 (step A).
4).

Next, the operation of the master-slave relationship information synchronization method in the distributed database system according to the first embodiment will be described in more detail with reference to FIGS.

(1) When the main table (master) 31 of the master server 1 is updated from the client 10 (see FIG. 4)

The main table (master) from the update processing means 11
31 (SQL; for example, inse
rt) to the RDBMS (master) 2.

'The RDBMS (master) 2 checks the validity of the update processing (SQL), and if there is an error in the requested update processing (SQL), the update processing means 11
To return an error.

The RDBMS (master) 2 checks the validity of the update process (SQL), and if the requested update process (SQL) is normal, registers the data in the main table (master) 31.

The RDBMS (master) 2 writes the contents updated in the update log file 4 as update log information (step A1).

The RDBMS (master) 2 notifies the RDBMS (slave) 8 of the updated contents.

The RDBMS (slave) 8 executes an update process (insert) on the main table (copy) 91.

The log monitoring unit 5 detects that the update log file 4 that has been regularly monitored has been updated (step A2).

The log monitoring means 5 calls the update reflection processing means 6 (Step A 3).

The update reflection processing means 6 acquires the latest update log information from the update log file 4.

The update reflection processing means 6 performs an update process on the main table (dummy) 93 using the acquired latest update log information (step A 4).

(2) When the client 10 updates the main table (copy) 91 of the slave server 7 (see FIG. 5).

The update processing means 11 is a main table (copy)
Data update processing (SQL; for example, inse
rt) to the RDBMS (slave) 8.

'The RDBMS (slave) 8 checks the validity of the update processing (SQL), and if there is an error in the requested update processing (SQL), the update processing means 1
Return an error to 1.

The RDBMS (slave) 8 checks the validity of the update process (SQL), and if the requested update process (SQL) is normal, registers the data in the main table (copy) 91.

The RDBMS (slave) 8 is an RDB
The MS (master) 2 is notified of the updated contents.

The RDBMS (master) 2 executes an update process (insert) on the main table (master) 31.

The RDBMS (master) 2 writes the contents updated in the update log file 4 as update log information (step A 1).

The log monitoring unit 5 detects that the update log file 4 that has been regularly monitored has been updated (step A2).

The log monitoring means 5 calls the update reflection processing means 6 (Step A 3).

The update reflection processing means 6 acquires the latest update log information from the update log file 4.

The update reflection processing means 6 performs an update process on the main table (dummy) 93 using the acquired latest update log information (step A 4).

(3) When the client 10 performs an update process on the master table 32 of the master server 1 (see FIG. 6).

The update processing means 11 is a slave table (master)
32 update processing (SQL; for example, inse
rt) to the RDBMS (master) 2.

'The RDBMS (master) 2 checks the validity of the requested update process (SQL) and the referential integrity constraint, and checks if there is an error in the update process (SQL) or the referential integrity constraint. If so, an error is returned to the update processing means 11.

The RDBMS (master) 2 checks the validity of the requested update process (SQL) and the referential integrity constraint, and if it is normal, the dependent table (master)
The data is registered with respect to 32.

The RDBMS (master) 2 is an RDBM
The S (slave) 8 is notified of the updated contents.

The RDBMS (slave) 8 executes an update process (insert) on the dependent table (copy) 92.

(4) When the client 10 updates the slave table 7 (copy) 92 of the slave server 7 (see FIG. 7).

The update processing means 11 is a slave table (copy)
Update processing (insert) of data to R.92
Request to S (slave) 8.

'The RDBMS (slave) 8 checks the validity of the requested update process (SQL) and the referential integrity constraint, and checks if there is an error in the update process (SQL) and the referential integrity constraint. If so, an error is returned to the update processing means 11.

The RDBMS (slave) 8 checks the validity of the requested update processing (SQL) and the referential integrity constraint, and if normal, registers the data in the dependent table (copy) 92.

The RDBMS (slave) 8 is an RDB
The MS (master) 2 is notified of the updated contents.

The RDBMS (master) 2 executes an update process (insert) on the subordinate table (master) 32.

In this embodiment, the master server 1
Although the case where two servers of the slave server 7 are used has been described as an example, the number of slave servers 7 may be plural.

If the slave server 7 is provided with the update log file 4, the log monitoring means 5 and the update reflection processing means 6, the roles of the master server 1 and the slave server 7 can be exchanged. be able to.

Next, a second embodiment of the present invention will be described.

FIG. 8 is a block diagram showing a configuration of a distributed database system to which a master-slave relationship information synchronization method according to a second embodiment of the present invention is applied.

Referring to FIG. 8, the second embodiment of the present invention has a configuration in which a master server 1 is provided with a recording medium 100 for recording a master server side program. The recording medium 100 may be a magnetic disk, a semiconductor memory, or another recording medium.

The master server program is stored in the recording medium 1
00, it is read into the master server 1 and operates as an RDBMS (master) 2, an update log file 4, a log monitoring unit 5, and an update reflection processing unit 6. Therefore, the operation of the master server 1 is the same as that of the first server described with reference to FIGS.
Since the operation is exactly the same as that of the master server 1 in the distributed database system to which the master-slave relationship information synchronization method according to the embodiment is applied, detailed description is omitted.

[0080]

The effect of the present invention is that the update process can be performed on the distributed master server while maintaining the master-slave relationship between the contents and the table only by updating the master table (master) of the master server. As a result, a process of updating data having the same content for each server is not required. The reason is that the update log information of the main table (master) is monitored by the log monitoring means, and only when the update is performed, the update contents are reflected on the main table (dummy) of the distributed slave servers. This is because

In addition, when a new subordinate table is added to the main table, conventionally, the subordinate table is defined (master server,
Although it was necessary to update data together with the slave server) and to implement consistency assurance on the application program side (server only), in the present invention, the definition of the slave table (master server and slave server) and the main table (dummy) ) Only needs to be set (to establish a master-slave relationship) for the foreign key, and there is an advantage that the slave table can be easily added.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a master-slave relationship information synchronization method in a distributed database system according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing an operation of a master-slave relationship information synchronization method in the distributed database system according to the present embodiment.

FIG. 3 is a diagram illustrating contents of relational data (master), relational database (slave), and update log file in FIG. 1;

FIG. 4 is an operation explanatory diagram in a case where an update process is performed on a main table (master) of a master server from a client in FIG. 1;

FIG. 5 is an explanatory diagram of an operation when an update process is performed on a main table (copy) of a slave server from a client in FIG. 1;

FIG. 6 is an operation explanatory diagram in a case where an update process is performed on a master table of a master server from a client in FIG. 1;

FIG. 7 is an operation explanatory diagram in a case where an update process is performed on a slave table (copy) of a slave server from a client in FIG. 1;

FIG. 8 is a block diagram showing a configuration of a master-slave relationship information synchronization method in a distributed database system according to a second embodiment of the present invention.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 master server 2 RDBMS (master) 3 relational database (master) 4 update log file 5 log monitoring means 6 update reflection processing means 7 slave server 8 RDBMS (slave) 9 relational database (master) 31 main table (master) 32 dependent table (Master) 91 main table (copy) 92 slave table (copy) 93 main table (dummy) 100 recording medium

Claims (6)

[Claims] 1. A relational database (master) in which a main table (master) and a subordinate table (master) are defined, and a master-slave relationship between the main table (master) and the subordinate table (master) is defined, and the relational database. A master server having an RDBMS (master) managing a database (master), a relational database (slave) in which a main table (copy) and a dependent table (copy) are defined, and an RDBMS (slave) managing the relational database (slave) In a distributed database system in which a slave server having ()) is connected via a line, a main table (dummy) is further defined and a master / slave table is provided between the main table (dummy) and the slave table (copy). The relational database (slave) in which the relationship is defined; The server,
An update log file in which update log information is accumulated when the main table (master) of the relational database (master) is updated by a DBMS (master), and an update log file provided in the master server, wherein the update log file is updated. Log monitoring means for periodically monitoring whether or not an update of the update log file has been detected by the log monitoring means based on the latest update log information of the update log file. And an update reflection processing means for reflecting the update contents to the main table (dummy) in (1). 2. The master table (master) and the slave table (master) are normalized, and the master table (copy) and the slave table (copy) are normalized. A master-slave relationship information synchronization method in the distributed database system described above. 3. The master-slave relationship information synchronization method in a distributed database system according to claim 1, wherein a plurality of slave servers are connected to one master server. 4. The RDBMS (master) does not violate a validity check of whether the update process is correct and a referential integrity constraint of a table when an update process of the slave table (master) is requested. It is checked whether it is normal, and if it is normal, the slave table (master) is updated, the RDBMS (slave) is notified of the update, and the RDBMS (slave) is updated from the main table (copy). 3) and the same update process is performed on the master table (master) and the slave table (master) when notified that the update process has been performed on the slave table (copy). A master-slave relationship information synchronization method in the distributed database system described above. 5. The RDBMS (slave) does not violate the validity check of whether the update processing is correct and the referential integrity constraint of the table when the update processing of the slave table (copy) is requested. It is checked whether it is normal or not, and if it is normal, the slave table (copy) is updated and the RDBMS (master) is notified of the update, and the main table (master) is notified from the RDBMS (master). ) And the same update processing is performed on the main table (copy) and the slave table (copy) when it is notified that the update processing has been performed on the slave table (master). A master-slave relationship information synchronization method in the distributed database system described above. 6. A log monitoring means for periodically monitoring whether or not the update log file has been updated,
And updating the main table (dummy) in the relational database (slave) based on the latest update log information of the update log file when the update of the update log file is detected by the log monitoring means. A recording medium on which a program for functioning as update reflection processing means for reflecting contents is recorded.