JP2000020374A - Replication control system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily reflect the original data base of a trunk host with updated contents at high speed in the case of executing an updating job through an open server by arranging the replicas of tables dividedly stored in the main host through plural open servers for each divided unit. SOLUTION: In the case of updating processing on open servers 1-3, the updating history of the data base is stored in an updating history storage means 106 and the updating history information stored in the updating history storage means 106 on the respective open servers 1-3 is collected into an updating history information storage means 108 of a trunk host 10 by an updating history centralizing means 107. Afterwards, an updating information reflecting means 109 for reflecting a data base storage means group 102 of the trunk host 10 with the updating history information collected into the updating history information storage means 108 is activated, a data base determining means 110 and a data base updating means 111 are operated and the original data base of the trunk host 10 is updated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a replication control system for a database system in a distributed environment.

[0002]

2. Description of the Related Art Conventionally, there is a distributed database function that uses a two-phase commitment protocol as a distributed database function for maintaining consistency of a database arranged in a distributed environment. When performing asynchronous update such as replication, place a replica of the original database at another site, and when the original database is updated, generate a replica update transaction from the log information and transfer it to the replica site for execution I am trying to do it. As a distributed database function for maintaining the consistency of a database arranged in a conventional distributed environment, for example, Japanese Patent Laid-Open No.
4718 and JP-A-7-219830.

[0003]

However, the conventional distributed database function has the following problems. The first problem is that the performance in the updating process is not good when performed by the distributed database function using the two-phase commitment protocol. The reason is that it is necessary to notify the database system of another site every time the update processing is performed and to confirm whether or not the update can be performed, thereby increasing the communication load.

[0004] The second problem is that when performing an asynchronous update with a replica at a different site from the original database,
In the method of reflecting the update of the original database on the replica, a database (original database) that divides one table on the main host is placed on multiple open servers (for example, divided and placed for each district), Update processing cannot be performed on the open server to reflect the original database. Also,
There is also a problem that a communication load is increased by updating the original database, generating a replica update transaction from the log information, transferring the transaction, and executing the transaction.

In the present invention, a database in which one table is divided on a main host is divided and arranged on a plurality of open servers, and updates made on the open server are reflected in an original database of the main host. To provide a replication control system that can collect update information on all open servers by the main host, instead of notifying the main host individually from the open server that performs the update. The purpose is.

[0006]

SUMMARY OF THE INVENTION A replication control system according to the present invention is a database arranged on a distributed system comprising a main host and a plurality of open servers, and a part of the database on the main host is copied. A replication control system that references / updates on an open server by performing a database definition unit that defines information required for performing replication to a database storage unit on each open server, and Update information creating means for creating update information when updated by the update request means, update history accumulating storage means for storing an update history, and storing the update history created by the update information creating means in the update history Update history storage means for storing in the storage means, and the update history on each open server Update history collecting means for collecting the update history information accumulated in the product storage means in the update history information storage means of the main host, and a group of database storage means of the main host for storing the update history information collected in the update history information storage means Update information reflecting means for determining the corresponding database storage means in the database storage means group based on the update information, and updating the contents of the determined database storage means. It is characterized by.

[Operation] In the replication control system, the database of the main host is divided and arranged in each open server in the distributed environment, and when the arranged database is updated, the update history is accumulated in the update history accumulating means. By starting a batch job that periodically collects update histories on all open servers from the main host, the update history collection unit operates and collects all update histories in the update history information storage unit of the main host. . Then, the database of the main host is updated based on the collected update histories of the divided databases on the open server.

[0008]

Next, an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of a replication control system according to an embodiment. The replication control system according to this embodiment includes a database definition unit 101 for defining a table in a database, a database storage unit group 102, and an update request unit 103 for updating the database storage unit group 102 from a program or the like on each open server. Update information creating means 104, update history accumulating means 105, update history accumulating and storing means 106, update history collecting means 107 for collecting the update history of each open server, update history information storing means 108, and core host 10. Update information reflecting means 109 for updating the database storage means group 102, a database determining means 110, and a data updating means 111.

Here, FIG. 2 shows a database definition means 1.
01 indicates the content defined by the table name 2
01 and replication primary key information 202
The division information 203 of the core host 10, the division information 204 of the open server, and the division information 20 of the core host 10.
3, a value / range 206, an open server division number 207 indicated by the open server division information 204, and a division number 208 of the core host 10.

FIG. 3 shows the update history information 301 stored in the update history storage means 106 and the update history information storage means 108. The resource information 302, the update type 303, and the record key value 304 And division number 3
05, pre-update information 306, and post-update information 307. The pre-update information 306 indicates a pre-update record format 308, and the post-update information 307 indicates a post-update record format 309.

Next, the operation of the replication control system according to this embodiment will be described. Prior to the start of operation of the database system, the user defines a replication table in the core host 10 and each open server by the database definition means 101. The contents to be defined are the contents shown in FIG. 2, and are a table name 201 of a table to be replicated, replication primary key information 202 which is information of a field for uniquely determining a record in the table, and table division information. The table division information differs between the table of the core host 10 and the table of the open server.
In the case of 0, the division information 203 of the main host 10 is defined, and a division number 205 of the table and a value / range 206 indicating a value or range at the time of division are set. In addition, each open server defines its own open server division information 204 in addition to the division information 203 of the core host 10, and has its own table division number 207 and division number 20 of the main host 10.
8 is set as a pair.

In the actual operation of the database system, each open server is in charge of a task within a divided range, and the database is referred to / updated. The database is used for the supervised business. Therefore, when the update (addition, change, deletion) of the database is performed in each open server, the update request unit 103 stores the database storage unit group 10 in its own.
2 is updated, update history information 301 is generated by the update information creation unit 104, and is stored in the update history storage unit 106 by the update history storage unit 105.

The accumulated update history information 301 includes resource information 301 for identifying a table name, an update type 303 indicating whether an update process is an addition, a change, or a deletion, and a replication for uniquely determining an updated record. Based on the record key value 304 of the primary key and its own division number, the main host 1
Division number 305 set by changing to the division number on 0
And pre-update information 306 and post-update information 307. The pre-update information 306 and the post-update information 307 differ depending on the update type. In the case of addition, only the post-update information 307 is provided.
In the case of deletion, only the pre-update information 306 is provided. Both information 3
06 and 307 are set only in the case of change.

When the database of each open server is updated in the above-described manner and the content differs from the database of the main host 10, the database storage means group of the main host 10 is replaced with the database storage means group of each open server. Must be identical. For this reason, the update history storage unit 1 on each open server is mainly composed of the core host 10.
In order to collect the update history information 301 stored in 06, the update history information collecting unit 107 is activated. The update history collecting unit 107 collects update history information 301 currently stored in the update history storage unit 106 from all open servers on the network, and stores the update history information 301 in the update history information storage unit 108. Then, update information reflecting means 1
In step 09, the update history information 301 of the update history information storage unit 108 is read, and the database determination unit 110 operates to determine the table to be updated and the divided area.
05 is determined. After the update target is determined, the data update unit 111 executes the pre-update information 306 and the post-update information 3 on the action corresponding to the update type 303 of the update history information 301.
07 to update the database.

[0015]

Next, an embodiment of the present invention will be described with reference to FIG. FIG. 4 shows the inventory management table 401 and the main host 1
0 Kanto area 402 and Open server 2 Kanto area 4
03, a database update AP 404, an update history accumulation storage means 405, update history information 406, an update history information storage means 407, and a Kanto area 408 after update reflection.

[Explanation of the operation of the embodiment] The database definition means 101 is provided in the core host 10 and each of the open servers 1 and 2.
Defines the replication primary key information 202 as an inventory management table 401 with a field name of a number, and the table is divided by a district code, and is composed of a Hokkaido district, a Tohoku district, a Hokuriku district, a Kanto district 402, and so on. Is done. In addition, the open server 2 has an inventory management table 4 for only the Kanto area 402 in the inventory management table of the core host.
03 is defined.

With the above configuration, the database update AP 404 operates on the open server 2 and the inventory management table 40
The stock quantity of record 2 of 3 is changed from “1000” to “300”
, The update history information 406 is accumulated in the update history accumulation storage means 405 by the update request means 103, the update information creation means 104, and the update history accumulation means 105. In the update history information 406, the resource information 302 is an inventory management table, the update type 303 is changed, the record key value 304 is 10, the division number 305 is 4 which is the division number of the main host, and the pre-update information 306 is "1000". Information 3 after update
07 becomes “300”. After that, the update history collecting means 107 collects the update history information of the update history accumulating and storing means of each of the open servers 1 and 2, so that the update history information 406 of the open server 2 is stored in the update history information storing means 407. You.

The update history information 406 stored in the update history information storage means 407 is read out by the update information reflection means 109, and the resource information 302 of the update history information 406 is determined by the database determination means 110 to be an inventory management table. It is determined that the table to be updated is the inventory management table 401, and the division number 305 of the update history information 406 is 4
Therefore, it is determined that the update history is the update history for the Kanto area 402 in the inventory management table 401. Then, the data updating means 111 updates the contents of the Kanto area 402 to the contents of the Kanto area 407 based on the update type 303 and the record key value 304 of the update history information 406, the information 306 before the update, and the information 307 after the update. It has the same contents as the inventory management table 403 of the open server 2.

[0019]

According to the present invention, the following effects can be obtained. The first effect is that, when a table divided and stored in the main host is operated between a plurality of open servers for each division unit or several division units, and a database update operation is performed on the open server side, This means that the update of the database of the main host can be reflected at high speed. The reason is that when the original and the replica are processed as before, using the two-phase commitment protocol or generating and transferring the update transaction on the replica side from the log information every time the original data is updated In contrast to the execution, there is a disadvantage that the communication load becomes high. On the other hand, in the present invention, the update history is stored in the update history storage means of each open server, and the entire processing is performed on the main host side. This is because, at the time of execution, the core host collects the data collectively, so that the communication load can be reduced.

The second effect is an improvement in operability. The reason is that instead of notifying from each open server, when a database in which one table is divided on the main host is divided and arranged on a plurality of open servers and update work is performed on each open server, the update is performed on each open server. That is, the collection processing is performed from the backbone host, which is the original database, and is reflected in the database.

[Brief description of the drawings]

FIG. 1 is an overall explanatory diagram of a replication control system according to an embodiment.

FIG. 2 is a diagram illustrating contents of a database definition unit in the replication control system according to the embodiment;

FIG. 3 is a diagram illustrating update history information in the replication control system according to the embodiment.

FIG. 4 is an explanatory diagram of a specific example of the replication control system according to the embodiment;

[Explanation of symbols]

 1, 2, 3 Open server 10 key host 101 database definition means 102 database storage means group 103 update request means 104 update information creation means 105 update history accumulation means 106 update history accumulation storage means 107 update history collection means 108 update history information storage Means 109 Update information reflecting means 110 Database determination means 111 Data updating means

Claims (1)

[Claims] 1. A database arranged on a distributed system comprising a main host and a plurality of open servers, wherein a part of the database on the main host is copied and referred / updated on the open server. A control system, which is a database definition unit that defines the information required to perform replication to the database storage unit on each open server, and is updated when the database is updated on the open server by the update request unit Update information creation means for creating information; update history accumulation storage means for storing an update history; update history accumulation means for storing the update history created by the update information creation means in the update history accumulation storage means; The update history information stored in the update history storage means on each open server is used as a main host. Update history collecting means for collecting the update history information in the update history information storage means, and update information reflecting means for reflecting the update history information collected in the update history information storage means in a group of database storage means of the core host, the update information A replication control system, wherein the reflection unit determines a corresponding database storage unit in the database storage unit group based on update information, and updates the content of the determined database storage unit.