JPH06119383A - Parallel distributed data base system - Google Patents

Abstract

PURPOSE:To shorten response time even in the case that plural processing requests to a data base system overlap each other by registering data to plural data base systems and processing plural accesses to the same data or the same data base simultaneously in parallel in the case that the frequency-in-use of the data is high. CONSTITUTION:When a data registering instruction is received, a registering data base managing means 112 adds a data base name and frequency-in-use information as additional information at the time when a data base system managing part 11 registers the data to the designated data base system. In the case that the frequency-in-use of the data is high, or in the case that it is expected that the data is accesses simultaneously with other data, it registers the same data to two or more data bases when a registering person gives an instruction. Accordingly, even in the case that the plural access requests come simultaneously to the same data, they can be processed simultaneously by the data base system to operate in parallel.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a parallel distributed database system which processes a plurality of access requests to a database in parallel and quickly retrieves necessary data.

[0002]

2. Description of the Related Art FIG. 6 shows the structure of a conventional database system disclosed in Japanese Patent Laid-Open No. 3-111948.
Reference numeral 1 is a management section communication means 611, a data file storage location management means 612, and a data file management information storage means 61.
3, distributed data file management means composed of data file access status monitoring means 614 and data file moving means 615, 62 is communication means 621, distributed data file input / output means composed of data file storage location search means 622 and data file access means 623 , 63 are communication means 631 and data file storage location search means 63
2 and a distributed data file input / output unit 633, a data file storage unit 64, 66, a distributed data file utilization unit 65, 67, and a communication line 68.

The distributed data file input / output means 62 and the data file storage means 64 constitute a database system, and the distributed data file input / output means 63 and the data file storage means 66 also constitute a database system. These database systems are communication lines. It is connected to the distributed data file management means 61 via 68.

In the conventional distributed database system having the above structure, data access is performed as follows. In the case of reading or rewriting data, by inquiring through the communication line 68 to the distributed data file management means 61 that centrally manages the data storage location,
The data file storage means 64, 66 in which the target data is stored are recognized, and the target data is obtained therefrom. In the case of data registration, the data is registered in the data file storage means 64 or 66 at a certain location, and the information of the location where the data is registered is stored in the distributed data file management means 61 via the communication line 68. To be done.

[0005]

Generally, it is desirable that the time from requesting data to the database system until receiving the data is short. However,
In the conventional database system, the same data is only one in the database system group connected by the communication line 68, and when access to the same data is duplicated, sequential processing is performed. Also, duplicate access to the same database system was similarly processed in sequence. Therefore, when a plurality of accesses to the same data occur at the same time, the process accessed later has to wait until the process accessed earlier ends. Further, when a plurality of accesses to different data in the same database system occur at the same time, the process accessed later has to wait until the process accessed earlier finishes.

The present invention has been made to solve the above problems, and provides a parallel distributed database system capable of shortening the response time even when two or more processing requests to the database system overlap. The purpose is to get.

[0007]

A parallel distributed database system according to the present invention comprises means for determining the frequency of use of registered data and causing one or a plurality of database systems to register the data according to the frequency of use. A means for storing which database system the data is registered is provided.

[0008]

According to the present invention, the frequency of use of registered data is determined, and if the frequency of use is high, the data is registered in a plurality of database systems, and the same data or a plurality of accesses to the same database system are executed in parallel. Processed at the same time.

[0009]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows the overall configuration of a parallel distributed database system according to this embodiment, and 11 is a database system management unit, which is a database system management means 111.
And a registration database management means 112, a usage frequency determination means 113, and a communication means 114. 1
2 to 14 are database systems, and database management means 121, 131 and 141, respectively.
Storage units 122, 132, 142 and communication means 123, 1
33 and 143, each database system 12-14 is connected to the database system management unit 11 by a communication line 15. Although three database systems 12 to 14 are provided, two or more database systems may be provided.

Next, the data read processing operation of the parallel distributed database system having the above configuration will be described with reference to the flowchart of FIG. First, in step 21, when the data read command is received, the database system management means 111 sends data to the database management means 121, 131, 141 via the communication means 114, the communication line 15 and the communication means 123, 133, 143. Inquire about the presence or absence. Upon receipt of this inquiry, each of the database management means 121, 131, 141 receives the inquiries from the storage sections 122, 132, 14
It is checked whether or not there is request data in 2, and the result is sent to the database system management means 11 via the communication means 123, 133, 143, the communication line 15 and the communication means 114.
Contact 1.

At step 22, the database system management means 111 determines the database management means 121, 131,
When receiving a response indicating that there is request data from one of 141, in step 23, the inquiry to the other database management means is canceled, and in step 24, the data is received. FIG. 4 shows information to be added to the actual data, 41 is the database name in which the actual data is registered, 42 is information on the frequency of use of the actual data, and 43 is the actual data. Then, of the received data, the additional information 41, 42 that is a part other than the actual data is passed to the registration database management means 112.

The registered database management means 112 determines the use frequency of this data from the use frequency information 42 of the data by using the use frequency determination means 113 in step 25, and determines whether or not the additional information needs to be updated in step 26. If necessary, the registered database name 41 and the usage frequency information 42 are rewritten in step 27, and the rewritten additional information is sent back to the database system management means 111. In step 28, the database system management means 111 registers the additional information 42 and the actual data 43 based on the registered database name 41 of the additional information sent back.

Next, the data rewriting processing operation will be described with reference to the flowchart of FIG. Upon receiving the data rewriting command, in step 31, the database system management means 111 checks the presence / absence of target data in the database management means 121, 131, 141 via the communication means 114, the communication line 15, and the communication means 123, 133, 143. As in the case of the inquiry and data reading, the inquiry to the database management means other than the database management means that first returned the response is canceled. The database management means that first responded receives the additional information 41 and 42 of the requested data from the registration database management means 112.
Contact.

In step 32, the registration database management means 112 recognizes from the received additional information which database system the same data is registered in, and in step 33 the database of all the database systems in which the target data is registered. Inquires the management means whether the data is being rewritten. The inquired database management means checks whether the requested data is being rewritten and informs the registered database management means 112. If it is confirmed that all the target data is not being rewritten, in step 34, the target data in the database system that is being rewritten is locked so as not to respond to other rewriting requests, and in step 35, the data is rewritten. Is rewritten.

When the rewriting is completed, in step 36, the registration database managing means 112 causes the usage frequency judging means 11 to operate.
3 is used to determine the data usage frequency from the data usage frequency information 42, and from this usage frequency it is determined in step 37 whether or not the additional information needs to be updated. And the usage frequency information 42 is updated. In step 39, the database system management means 111 updates the updated actual data 43 and additional information 4
1, 42 are registered according to the registration database name 41, and the data lock is released in step 310.

Next, the data registration processing operation will be described. Upon receiving the data registration command, the database system management means 111 registers the data in the designated database system. At that time, the registration database management means 1
12, a database name 41 registered as additional information and usage frequency information 42 are added. If the data is used frequently or is expected to be accessed at the same time as other data, when the registrant indicates so, the same data is registered in two or more database systems.

By registering the same data in a plurality of databases as described above, even when a plurality of access requests are made to the same data at the same time, they can be processed simultaneously by a plurality of database systems operating in parallel. Further, even when a certain database system wants to access other data in the database system during request processing, since the same data is registered in the other database system, it can be processed without waiting.

FIG. 5 shows a more specific structure of the parallel distributed database system described above. Data A is registered in the storage units 122 and 132 of the database systems 12 and 13, and data B is stored in the storage units of the database systems 12 and 14. The data C is registered in the storage units 132 and 142 of the database systems 13 and 14, respectively. Here, when access requests are simultaneously sent to the data A, B, and C, it is assumed that the database system 12 first responds to the data A, and the data A is processed by the database system 12.

Next, for data B, database system 12 is processing for data A, so database system 14 reacts faster. Therefore, the data B is processed by the database system 14. In this way, multiple accesses are processed simultaneously. Similarly, even when two access requests are made to the data A at the same time, since the data A is registered in the database systems 12 and 13, the two accesses can be processed at the same time.

[0020]

As described above, according to the present invention, in the case of frequently used data, the same data is registered in a plurality of independently operating database systems. Even if they are duplicated in the same database system, they can be simultaneously processed in parallel, and therefore the response time to the database system can be shortened. Further, even if one of the plurality of database systems fails, the processing can be continued as it is.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a parallel distributed database system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a data read process of the parallel distributed database system of FIG.

3 is a flowchart showing a data rewriting process of the parallel distributed database system of FIG.

4 shows information added to data in the parallel / distributed database system of FIG.

5 is a more specific configuration diagram of the parallel distributed database system of FIG.

FIG. 6 is a configuration diagram of a conventional distributed database system.

[Explanation of symbols]

 11 database system management unit 12 to 14 database system 15 communication line 111 database system management means 112 registered database management means 113 use frequency determination means

Claims (1)

[Claims] 1. A multi-database system including a plurality of database systems that register and manage data and operate independently, and a database system management unit that is connected to each database system by a communication line and manages each database system. A means for determining the frequency of use of registered data, and registering the data in one or a plurality of database systems according to the frequency of use, and a means for storing in which database system the data is registered A parallel distributed database system comprising: