JPH06274401A - Distributed data base control system - Google Patents

Abstract

PURPOSE:To improve the response performance by registering its data query, and setting it to a temporary cache, when a data base on a remote node is accessed. CONSTITUTION:When a query is issued from an application program 1, a query analizing means 21 analyzes the query, and checks if there is one which includes the current query from in the query which is executed and held already. As a result, in the case the included existing query does not exist, information of the query is held in a query holding means 22. On the other hand, in the case included existing query exists, it is rewritten to a temporary table name being a cache of the existing query which proved to be included. Also, the latest access time of the reutilized query is updated to the current time. Subsequently, a data management system 3 executes an access of a data base in accordance with a location node of a table.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a relational database access request issued by an application program on a computer network capable of transmitting and receiving data between computers by a communication means, and a remote computer for data not existing in its own node. The present invention relates to a distributed database control method in a database management system capable of accessing the above database, and particularly to a distributed database control method in which an access request and an access result are stored as a pair and reused at a later request.

[0002]

2. Description of the Related Art Conventionally, in this type of distributed database control method, the management of the data obtained from the remote database is left to the application program, and the data is removed from the control of the database management system or the limited data is controlled. It was limited to whether it was downloaded periodically and operated as a local database.

[0003]

The above-mentioned conventional distributed database control method is used in the case where the access execution result, the data acquired from the remote database already exists in the local node, and the result of the next query is included. However, there is a drawback that the remote database access process is executed again for the next query. Further, in order to avoid this situation, a burden occurs such that the application program must manage the distributed data independently.

Next, when download is possible, the above-mentioned problem of duplicated costs is somewhat improved, but it is generally difficult to adapt although it is effective for a standard access pattern that can be predicted in advance. It is also difficult to design what to download at what timing.
Further, there is a drawback that the network is expanded and real-time property is required under the condition of 24H operation, and basically it becomes difficult to download itself.

[0005]

DISCLOSURE OF THE INVENTION The distributed database control method of the present invention comprises a query analysis means for analyzing a query issued by an application program and a query holding means for holding the query for later access. And a query rewriting means for rewriting the query so that if there is an executed query including the query, the access to the remote database is supposed to be the data cached in the own node. A query storage unit including: a query storage unit configured to generate a name capable of identifying a query held by the query storage unit in a one-to-one correspondence with data obtained by accessing a remote database; Temporarily store the data in the database management system with the name obtained by the associating means. The registration request is a as a cache Buru and a data cache unit having a cache registration unit.

Further, in the distributed database control method of the present invention, the data cache means, when the total amount exceeds a certain capacity at the time of registering the cache, sets the generation order of each query held by the query holding means to the query rewriting means. Cache replacement means for deleting the oldest cache and the corresponding query and registering the latest cache based on the correction order that is sequentially updated so that the order of the query corresponding to the cache to be reused becomes the latest.

[0007]

The present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the functional arrangement of a distributed database control system according to an embodiment of the present invention. Also,
FIG. 2 is an explanatory diagram showing a query holding format. In FIG. 1, the distributed database control system of the present invention is composed of an application program 1, a query storage unit 2, a database management system 3, a communication unit 4, and a data cache unit 5. The query storage unit 2 includes a query analysis unit 21, a query holding unit 22, and a query rewriting unit 23. The data cache unit 5 includes a query association unit 51, a cache registration unit 52, and a cache replacement unit 53 (corresponding to claim 2). Of course, there may be an example in which the cache replacement unit 53 is not included.

Referring to FIG. 2, the query holding format shows the correspondence between the query issued from the application program 1 and the temporary table name for caching remote data as the execution result. The date and time of the most recent access, the actual expression of the query, the name of the table and the node name where the query is located (obtained from the data dictionary or system catalog, etc., if necessary)
And the names of columns (projected columns and columns included in constraints) included in the query, and the name of the temporary table to be saved as a cache (in this example, the process id of the application program, the sequence number within the process, and the suffix tmpptbl). (Represented by and)) and five columns. The query holding unit 22 writes the first four items, and the query rewriting unit 2
According to 3, the first item is corrected when the cache is used.
The fifth item is determined by the query associating unit 51.

However, the latest access date and time of the first item in the above query holding format is not always necessary,
It is also possible to represent by link structure instead of date and time (corresponding to claim 2).

FIG. 3 is a flow chart showing the operation of the above distributed database control method.

First, how a query is stored for queries originating from the application program,
Or, an existing query execution result will be used or a mechanism thereof will be described (query storage means 2), and then a mechanism of access result data cached corresponding to the query will be described (data cache means 5). .

When a query is issued from the application program 1, the query analysis means 21 analyzes the query and determines whether or not there is a query that includes the current query among the queries that have already been executed and saved. Is checked (steps 101 and 102).

For general determination of formal inclusion, see A. K
For example, the method for lug tableau is used. However, it is determined that there is no inclusion in cases that cannot be determined, such as when the columns required for query execution are not cached. There is a case where the table location site name is not explicitly specified in the table included in the query, but at that time, the site name is clarified by checking the system catalog or data dictionary.

Next, when there is no existing query to be included, the query holding unit 22 saves the query information in the format shown in FIG. 2 (step 103).
If there is an existing query to be included, the query rewriting means 23 replaces the table name in the current query with a temporary table name which is a cache of the existing query that is found to be included (step 10).
4). Furthermore, the latest access date / time of the reused query is updated to the current date / time. The query that has undergone the above processing is requested to the database management system 3 (step 105). The database management system 3 uses the communication means 4 to access the database according to the node where the table is located, if necessary.

Next, the mechanism of the cache of the data obtained as a result of access will be described.

The execution result processed by the database management system 3 is returned to the application program (step 106), and at the same time, if the current query is held, cache processing is performed (step 107 and subsequent steps). The processing order of both depends on the implementation method, but in this example, the processing is performed in parallel in consideration of the response performance.

First, the query associating means 51 names the temporary table which is the fifth item of the query holding format. In this example, the process id of the application program, the serial number as the process unique number, and the suffix t
Although it is named as mptbl, the naming method is arbitrary as long as no collision finally occurs. Next, the cache registration means 52 requests the database management system 3 to register a temporary table as a cache as in the case of normal table definition / data registration.

If the cache registration unit 52 is allowed to register the cache indefinitely, the area of the database used for the original purpose may be squeezed. Therefore,
The cache replacement unit 53 efficiently implements a measure that sets a certain allowable amount and does not use it for the purpose of cache beyond that. The result data amount of the current query is added to the total amount of cache, and if it is less than the allowable amount, it is registered as it is (step 109). Information and cache data are deleted (step 108). Repeat this operation as many times as necessary to register the current cache data.

When the link structure is adopted without using the date and time, in order to guarantee the access order, it is necessary to relink the link at the timing of data access and cache replacement.

[0021]

As described above, according to the present invention, once the database on the remote node is accessed, the data and the query are registered in the database of the own node in the same manner as a normal table, and the temporary table is temporarily stored. Keep it in the cache. When the next access is included in the cache, it is not necessary to access the remote node again. That is, the application program can be realized without being aware of it.

Further, the present invention has an effect that it can be adapted even in the case where it is difficult to download in an atypical business form or an operating form of 24H operation on a large-scale network.

Further, by setting the cache allowable amount and replacing the oldest used one, there is an effect that efficient operation can be performed without putting pressure on the general database use area.

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional configuration of an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing a query holding format.

3 is a flowchart showing the operation of the embodiment of FIG.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Application program 2 Query storage means 3 Database management system 4 Communication means 5 Data cache means 21 Query analysis means 22 Query holding means 23 Query rewriting means 51 Query associating means 52 Cache registering means 53 Cache replacing means

Claims (2)

[Claims] 1. A database on a remote computer for data that does not exist in its own node in response to a query, which is a relational database access request issued by an application program on a computer network capable of receiving data between computers by means of communication means. In a distributed database control method relating to a database management system capable of accessing a query, a query analysis unit that analyzes a query issued by an application program and has already executed, a query holding unit that holds the query for later access, and the query And a query rewriting means for rewriting the query so that the access to the data that would otherwise be accessed on the remote database is cached in the local node when there is an executed query including A query storing unit having the query storing unit, a query associating unit for generating a name capable of identifying the query held by the query holding unit with respect to the data obtained by accessing the remote database in a one-to-one correspondence, and the query associating unit. A distributed database control method, comprising: a data cache unit having a cache registration unit for making a registration request to the database management system as a cache which is a temporary table with the name obtained by the unit. 2. The data cache unit, when the total amount exceeds a certain capacity at the time of registering the cache, the query corresponding to the cache that the query rewriting unit reuses the order of occurrence of each query held by the query holding unit. 2. The distributed database control method according to claim 1, further comprising cache replacement means for deleting the oldest cache and the corresponding query and registering the newest cache based on the correction order that is sequentially updated so that the latest order is updated.