JP3527834B2 - Distributed database system - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed database system, and more particularly, to a CSS (Client Server System).
The present invention relates to a method for controlling switching of a used database in a distributed database system when a system for executing business processing in m) is constructed. 2. Description of the Related Art In recent years, it has been common practice to construct a system for executing business processes by using a CSS using a distributed database system. FIG. 2 is an explanatory diagram schematically showing a conventional system construction example. The business process is divided into a client-side business process 201 and a server-side business process 202, each of which is operated by a client machine and a server machine. The server machine includes a database management system (hereinafter, referred to as DBMS) 203 and a server database (hereinafter, referred to as server DB). For example, in the server DB 204, there are tables A, B, and C that are commonly used in the entire system, and a part of the data is extracted from the tables A, B, and C and created. When performing a task of editing and processing data on the table T, the table T is generated in the server DB 204.
Each time access is made, communication between the client and the server occurs. In particular, if access to the table T occurs frequently, communication overhead between the client and the server increases. When a plurality of clients access the table T, the load is concentrated on the server machine. On the other hand, when a system using a database is constructed, specific rows and columns are extracted from a plurality of tables, created and edited, processed, and then deleted at the end of a transaction. It is common practice to provide a generic table (hereinafter referred to as a temporary table). At this time,
If the temporary table is on the server machine, the load is concentrated on the server machine as the number of clients increases. As described above, when constructing a system for executing business processing by CSS, there is a problem that the load is concentrated on the server machine as the number of clients increases. Therefore, it is being devised to distribute the load of the server machine. Such a technique is described in, for example, JP-A-6-259308 and JP-A-8-259308.
-339323. [0008] On the other hand, there is a case in which a target business process is performed on a client machine that is free at the time of operation. In this case, the business process operating on the client machine changes dynamically. [0009] Among the above-mentioned prior arts, in the technique disclosed in Japanese Patent Application Laid-Open No. 8-339323, if the number of accesses exceeds a certain number in a plurality of server machines, By moving data (distributed data) first, the arrangement of the distributed data is optimized. However, the technique disclosed in the prior application generates all tables on the server machine.
The communication overhead between the client and the server for the temporary table cannot be suppressed. In the technique disclosed in JP-A-6-259308, the load on the server machine and the availability of the storage medium are monitored, and the load and the availability of the storage medium are balanced. However, according to the technology disclosed in the prior application, necessary data exists in a remote machine depending on the load and the availability of a storage medium, and communication overhead for data access increases. Things happen. In particular, if a table, such as a temporary table, that needs to exist only for one transaction exists on a remote machine, unnecessary communication overhead occurs. The present invention has been made in view of the above points,
The purpose is to generate a table for storing data extracted from a plurality of tables of the server machine on each client machine, and when business processing is executed on the client machine, the table is added to the above table. By storing the data extracted from the plurality of tables in the above, and thereafter completing the processing in the client machine, it is possible to edit and process the data in the above table in the client machine. An object of the present invention is to reduce the communication overhead between machines and to distribute the load on the DBMS of the server machine. In particular, in the present invention, the definer considers that the efficiency is highest when the business process to be executed on each client machine is present in the client machine that performs the business process, such as a temporary table in an operation mode that dynamically changes. An object of the present invention is to suppress unnecessary communication overhead by always placing a recognizable table on a client machine that performs business processing. Further, another object of the present invention is to achieve the above object by identifying information for identifying whether or not to generate a database table on a client machine.
When the definer changes, the system automatically creates and deletes tables on each client machine,
An object of the present invention is to reduce the troublesome operation of the definer. [0015] To achieve the above object, a distributed database system according to the present invention comprises:
In addition to the database under the server machine, a database is provided under each client machine, and means for identifying identification information as to whether or not the processing target table is a table generated on each client machine is provided. Then, at the time of definition, the definer specifies, for each table, identification information (type) indicating whether or not the processing target table is a table to be generated on the client machine, thereby automatically generating a table to be generated for each client machine. Generated in the database under the client machine, at the time of operation,
Data to be stored in the table generated in each client machine is extracted from the database table under the server machine, stored, and then accessed to the database under business processing under the client machine. In addition, for a table generated in each client machine, a table for managing information necessary for generating this table is provided, and means for making a request to generate a table in the DBMS of each client machine is provided. When the definer changes the identification information indicating whether or not the table is generated on each client machine, the system automatically generates and deletes the table on each client machine. By doing so, the database to be accessed can be determined by identifying the identification information as to whether or not the processing target table is a table generated in each client machine. If the table is generated on each client machine, access to the server machine can be suppressed. Further, with respect to a table generated in each client machine, a table for managing information necessary for generating this table can efficiently generate a request for generating a table. Further, when the definer changes the identification information indicating whether or not the table is generated in each client machine, the means for making a request to generate a table in the DBMS of each client machine is necessary for generating the table. Since a request for generating a table can be transmitted to each client machine from the table for managing information, manual operation by the definer is not required. Embodiments of the present invention will be described below with reference to the drawings. FIG. 3 is a diagram schematically showing the configuration of the distributed database system according to one embodiment of the present invention. As shown in FIG. 3, the business process is divided into a client-side business process 301 and a server-side business process 302, each of which is operated by a client machine and a server machine. The server machine 303 has a server DBMS 303
And a server DB 304. The client machine also has a client DBMS 305 and a client database (hereinafter, referred to as a client DB) 306.
And exists. Here, the client DB 306 is a database that stores tables generated in each client machine, and the client DBMS 305 is a database management system that manages tables in the client DB 306. Since a table generated in each client machine is stored in the client DB 306, when accessing this table, communication between the client and the server is not required, and the load conventionally concentrated on the server DBMS 303 is reduced. , Can be distributed to the client DBMS 305. For example, in the server DB 304, there are tables A, B, and C that are commonly used in the entire system, and tables created by extracting some data from the tables A, B, and C are created. In the case of performing a task of frequently editing and processing data on the table T in the T, the table T is generated in the client DB 306, and necessary data is obtained from the tables A, B, and C of the server DB 304, and the table T is obtained. By storing and editing and processing the data of the table T, communication between the client and the server becomes unnecessary, and the load of the server DBMS 303 required for the processing of editing and processing the data of the table T is reduced by the client DBM.
It can be distributed to S305. FIG. 1 is a configuration diagram of the distributed database system of the present embodiment. In the figure, 10 is a client machine, 11 is a client DBMS, 108 is a client DB, 20 is a server machine, 21 is a server DBMS, 104 is a server DB, and 120 is a shared disk. First, the operation at the time of definition will be described. The table definition unit 101 of the client machine 10 is a part that defines components of the database such as tables and indexes according to the instructions of the definer. The definer is the client machine 10
When a table or an index is newly defined, the request is sent to the table definition unit 1 of the server machine 20 via the LAN 102.
Send to 03. The table definition unit 103 of the server machine 20 updates the definition information 105 of the server DB 104 via the DBMS kernel unit of the server DBMS 21, generates a table or index defined by the definer, and then generates a table distribution management unit 10
Pass control to 6. FIG. 4 shows a table management table which is a part of the definition information 105. The table management table includes type information 401 and generation date 402 in addition to the identifier and the table name. Type information 401
Stores the identification information of whether or not the table is generated in each client machine 10 for each table, and the generation date and time 402 stores the date and time when the table was generated. When the value of the type information 401 is "client", it indicates that the table is a table generated for each client machine, and the value of the type information 401 is "server".
Indicates that the table is generated only on the server machine. When the value of the record type information 401 of the newly added / updated table management table is “server”, the table distribution management unit 106 immediately transmits the result of the change to the server DB 104 to the table definition of the client machine 10. To the unit 101. The table distribution management unit 106 also stores the type information 4
When the value of “01” is “client”, the definition information 10
5 is a script that updates the definition generation management table, which is a part of the table 5, updates the file of the request destination list 107 in the shared disk 120, and generates a table, an index, and the like defined by the definer for each client machine 10. Send a sentence.
The detailed operation of the table distribution management unit 106 will be described later. FIG. 5 shows a definition generation management table which is a part of the definition information 105. The definition generation management table includes a table name 501,
There are a table generation statement 502, an index generation statement 503, and the like. The table name 501 stores the table name of the table defined by the definer, the table generation statement 502 stores a script statement for generating the table, and the index generation statement 503 stores the script for generating the index of the table. Stores a statement. Each client machine 10 executes a script sent from the table distribution management unit 106 of the server machine 20 to generate a table, an index, and the like defined by the definer. Thereby, the client DB 108
In addition, a table defined by the definer can be generated. FIG. 6 shows the contents of the request destination list 107 of the shared disk 120. The request destination list 107 includes a table name 601 and a generation date / time 602, and the table distribution management unit 106 stores information on the table name and generation date / time of the table management table. Next, the operation during operation will be described. When the client-side business process 109 generates an access to the database, the request destination determination unit 110
A request destination list exists in the client DB 108 by referring to the request destination list 107 in the server DB 104.
To determine if it exists. The detailed operation of the request destination determination unit 110 will be described later. When the request destination table exists in the client DB 108, the client DB 108 accesses the client DB 108 via an inquiry processing unit and a DBMS kernel unit in the client DBMS 11 of the client machine 10, and
When the table of the request destination exists in the server DB 104, an inquiry processing unit in the server DBMS 21 of the server machine 20;
The server DB 104 is accessed via the DBMS kernel. Here, the request destination table is generated by a client machine other than the currently operating client machine. At this time, the request destination table is stored in the client DB 108 because the client machine 10 is stopped. Even if it is determined that it exists, the client DB 10 of the client machine 10
8 does not exist, or even if it exists, the definition contents may be different. If it is determined that the requested table exists in the client DB 108 but does not exist, the corresponding record is taken out from the definition generation management table of the definition information 105 of the server DB 104 and executed. , Generates a request destination table in the client DB 108. Further, even if it exists in the client DB 108, it is determined whether or not the definition content is the latest, based on the generation date 402 in the table management table in the client DB 108,
The comparison is made with the generation date and time in the file of the request destination list 107. Next, the operation of the table distribution management unit 106 will be described. After the generation of the processing target table, the generation of the index of the processing target table, and the update of the table management table are completed, the table distribution management unit 106
Performs the processing. FIG. 7 is a processing flowchart showing the flow of the operation of the table distribution management unit 106. First, the table distribution management unit 106
First, create the definition generation management table records such as the table name of the processing target table, information identifying the client machine of the request source, the table generation statement of the processing target table, and the index generation statement of the processing target table. And give it the necessary information to start execution. Here, the generation of the processing target table, the generation of the index of the processing target table, and the updating of the table management table are performed before the processing by the table distribution management unit 106 is executed. The generation statement, the index generation statement of the processing target table, and the like have already been obtained and can be given to the table distribution management unit 106. The table distribution management unit 106 is a server DB 104
Is searched by the table name of the given processing target table (step 701). As described above, since the update of the table management table has been completed, the corresponding record is always found. Therefore, the value of the type information 401 in the record of the found table management table is evaluated (step 702). If the value of the type information 401 is "server",
The record of the file of the request destination list 107 in the shared disk 120 is searched by the given table name (step 71).
3) Evaluate the existence of the record (step 71)
4). If the corresponding record exists in the file of the request destination list 107 in the shared disk 102, the corresponding record of the file of the request destination list 107 is deleted (step 715). Thereafter, the execution result is transmitted to the requesting client machine 10 (712), and the table distribution management unit 106 ends the processing. If the value of the type 401 is "client", the definition generation management table of the server DB 104 is searched by the table name of the given processing target table (step 703). Thereafter, it is evaluated whether or not the corresponding record exists in the definition generation management table (step 704). If the relevant record exists in the definition generation management table, use the information required to create the record of the definition generation management table, such as the table generation statement of the given processing target table and the index generation statement of the processing target table. Update the corresponding record in the definition generation management table (step 70)
5). On the other hand, if the corresponding record does not exist in the definition generation management table, it is necessary to create a record of the definition generation management table such as a given table generation statement of the processing target table and an index generation statement of the processing target table. Using the necessary information, a corresponding record of the definition generation management table is newly added (step 70).
6). Next, the record of the file in the request destination list 107 in the shared disk 120 is searched by the given table name (step 707), and the existence of the record is evaluated (step 708). Request destination list 10 in shared disk 120
7, the corresponding record exists in the file of the request destination list 107, and the value of the item of the creation date and time 602 of the corresponding record in the file of the request destination list 107 is determined using the creation date and time in the corresponding record of the table management table found in step 701 It is changed (step 709). On the other hand, if the corresponding record does not exist in the file of the request destination list 107 in the shared disk 120, the generation date and time in the corresponding record of the table management table found in step 701 is used. A record is newly added to the file (step 710). Thus, in the file of the request destination list 107 in the shared disk 120, a record for a table whose type information 401 value in the record of the latest table management table is “client” is included. . After that, for all the currently connected client machines 10, the table generation statement and the index generation statement are transmitted together with the generation date and time on the server machine (step 711). Then, the requesting client machine 10
To the table distribution management unit 106 (712).
Ends the processing. When the client machine 10 receives the above-mentioned transmission message, a time later than the generation date and time at the server machine 20 (the current time at the client machine 10 at that time) is received. If the time is later than the generation time, the time is not changed,
If it is earlier than the generation time on the server machine that sent it, change the current time to the generation date and time on the server machine that sent it), and then execute the table generation statement and index generation statement . By comparing the generation date and time of the table management table in the client DB 108 with the generation date and time in the request destination list 107 in the shared disk 120, it is determined whether the table in the client DB 108 is the latest table. Can be determined. Next, the operation when the request destination determination unit 110 is operated by giving the request destination table name will be described. FIG. 8 is a processing flow chart showing the flow of the operation of the request destination determination unit 110. First, the request destination determination unit 110
Searches for the file of the request destination list 107 in the shared disk 102 with the given table name (step 801), and evaluates whether there is a corresponding record (step 802). If there is no corresponding record, the request destination table is stored as a record whose type information 401 is "server" in the table management table in the server DB 104. Table is Client DB10
8 is known not to exist. Therefore, the request destination table is a table to be accessed to the server DB 104 (step 813). Conversely, if the corresponding record exists, the request destination table is a table management table in the server DB 104, and the type information 4
01 is stored as a record indicating “client”, the request destination table is stored in the client DB 108.
Then, a table that should exist. However, at this time, the request destination table does not always exist with the latest definition in the client DB 108, so that a check needs to be performed. Therefore, the generation date and time of the record in the file of the request destination list 107 in the shared disk 120 found in step 801 is set to the variable FDate (step 803). Next, the table management table in the client DB 108 is searched with the given table name (step 804), and the presence or absence of the corresponding record is evaluated (step 805). If the corresponding record exists, the variable Ddate
The value of the generation date and time in the corresponding record is stored in (step 806). Then, the value of the variable Fdate set in step 803 and the variable Ddate set in step 806
(Step 807). Here, if the value of the variable Fdate is newer, it can be determined that the request destination table has not been generated with the latest definition in the client DB 108, as in the case where the corresponding record does not exist in step 805. . Conversely, if the value of the variable Fdate is not newer, the request destination table is
Since the request destination table has been generated with the latest definition, the request destination table is a table to be accessed by the client DB 108 (step 814). If it is determined that the request destination table is the client DB 108 and that it has not been generated with the latest definition, it requests the server DB 104 for a table / index generation statement and its generation date and time information ( Step 80
8). Next, the generation date and time returned from the server DB 104 is compared with the current time of the client machine (step 809). The current time with the date and time of generation on the server machine (step 81
0). Thereafter, the table generation statement and the index generation statement returned from the server DB 104 are executed (step 8).
11). The request destination table is now in the client DB 108
In, the request destination table is a table to be accessed by the client DB 108 (step 812). As is clear from the description of FIG. 8, the definition of the table generated in the client machine 10 can be automatically unified by the system. As described above, according to this embodiment, the table T is conventionally generated in the server DB 204 as shown in FIG. 2, whereas the table T is stored in the client DB 3 as shown in FIG.
06, communication between the client and the server is not required to access the table T. Further, thereby, the load of the server DBMS 303 can be distributed to the client DBMS 305. Further, the system can automatically generate and delete tables in each client machine only by changing the value of the type information 401 of the table management table. As described above, according to the present invention, in a distributed database system in which a server machine and a plurality of client machines connected to the server machine execute business processing,
When each client machine processes data that is handled uniquely by the client machine when processing business,
By completing the process only in the client machine,
It is possible to reduce the communication overhead between the server machine and the client machine and to distribute the DBMS load on the server machine. This is particularly effective when a client machine executes processing for a table such as a temporary table that does not permanently occupy a storage medium. Further, if the definer changes the identification information as to whether the database table is a table for storing data uniquely handled by each client machine or a table commonly handled by all client machines, the system By automatically generating / deleting a table on each client machine by itself, it becomes unnecessary for the definer to perform a special manual operation, and the trouble of the definer can be greatly reduced.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a configuration diagram of a distributed database system according to an embodiment of the present invention. FIG. 2 is an explanatory diagram schematically showing a conventional system. FIG. 3 is an explanatory diagram schematically showing a distributed database system according to one embodiment of the present invention. FIG. 4 is an explanatory diagram showing a table management table which is a part of definition information in one embodiment of the present invention. FIG. 5 is an explanatory diagram showing a definition generation management table which is a part of the definition information according to the embodiment of the present invention. FIG. 6 is an explanatory diagram showing file contents of a request destination list in one embodiment of the present invention. FIG. 7 is a processing flowchart showing a flow of operation of a table distribution management unit according to the embodiment of the present invention. FIG. 8 is a process flowchart showing an operation flow of a request destination determination unit in one embodiment of the present invention. [Description of Signs] 10 Client machine 11, 305 Client DBMS 20 Server machine 21, 303 Server DBMS 101 Client machine table definition unit 102 LAN 103 Server machine table definition unit 104, 304 Server DB 105 Server DB definition information 106 Table Distribution management unit 107 Request destination list 108, 306 Client DB 109 Client side business process 110 Request destination determination unit 111 Server side business process 120 Shared disk

Continuation of the front page (56) References JP-A-7-334402 (JP, A) Keller A. M. et al. , A Predicate-based Caching Scheme for Client-Server Database Architectures, VLDB Journal, Germany, Springer, 1996, Vol. 5, No. 1, p. 35-47, URL, http: // www. info matik. uni-trier. de / ~ ley / db / journals / vldb / KellerB96. html Dar S.H. et al. , Semantique Data Caching and Replacement, Proceedings of 22th International Conference on Very Large Data Bases, Morgan Kaufmann, 1996, p. 330-341, URL, http: // www. infomatik. un i-trier. De / ~ ley / db / conf / vldb / DarFJST 96. html (58) Field surveyed (Int. Cl. ⁷ , DB name) G06F 12/00 G06F 17/30

Claims (1)

(57) [Claims] [Claim 1] A server machine having a database,
Multiple connected to it, each with a database
Distributed data for executing business processes on client machines
In the base system, shares connected to the server machine and each client machine
A disk is provided, and this shared disk has a server machine
Each client machine in the table management table created by
Table names for tables that exist in each database
The request destination list information including the generation date and time is stored, and the client machine requests the shared disk for the request destination list information.
The request destination table is referenced from the
Database on the client machine
Is determined to exist in the server machine, and the request destination table is
Database, if it exists in the database
Access the base to get necessary data and client
Generates a table in the database of the remote machine, and
Exists in the client machine database and the request destination table is
If it is generated with the latest definition,
Access tables already generated in Thin's database
Table to be requested and the request destination table is the data of the client machine
The requested table is generated with the latest definition even if it exists in the base.
If not, access the database on the server machine.
And obtain the necessary data to
A table is created in the database, and each client machine processes
Client machine in client machine database
Table for data storage for exclusive use by
System automatically creates the
Manipulate table data for exclusive use in Thin
Access to the server machine
And a distributed database system.