Database system and a method of data retrieval from the system

Abstract

Since metadata pertaining to real data stored in at least one database (DB) are collected and managed at a single meta DB server, and metadata that match a retrieval request are extracted by search of the meta DB server, even when a plurality of DBs and DB servers for managing DBs are present on a network, all metadata that match the retrieval request can be extracted independently of which DBs metadata pertain to. Hence, all data that match a retrieval request can be obtained from a single server independently of the actual locations of the distributed DBs and DB servers.

Description

BACKGROUND OF THE INVENTION [0001] 1. Field of the Invention [0002] The present invention relates to a database system, data retrieval method, and storage medium and, more particularly, to a technique suitably used in a retrieval system that finds out desired data from a plurality of distributed databases. [0003] 2. Description of the Related Art [0004] As the performance of computers becomes higher in recent years, a large-scale computer such as a single main frame or the like has been replaced by a distributed system built by a plurality of workstations or personal computers in recent years. The distributed system makes development and maintenance of the system relatively easy. As an example of the distributed system, the so-called Internet is known. [0005] In the Internet, a plurality of computers are distributed worldwide as servers or clients, and construct a single, huge database (to be abbreviated as a DB hereinafter). Text information, image information, and the like are reg...

AI Technical Summary

Benefits of technology

[0022] It is an object of the present invention to provide a mechanism that allows the user to easily search a DB system built by distributed DBs and their servers without requiring immediate connectivity to the distributed DB servers.

[0023] It is another object of the present invention to provide a mechanism which always recognizes information pertaining to each DB stored in distributed servers, and allows the user to retrieve the latest information at the time of search as a result without exerting any extra load on the user.

[0024] It is still another object of the present invention to provide a database system which can join tables at high speed with a minimum required memory capacity in a relational database system.

[0040] According to still another feature of the present invention, since the second server comprises the function of retrieving metadata when the first server has updated metadata or at given time intervals, metadata collected at the second server allow the user to always obtain the latest information at the time of search as a result, thus flexibly coping with changes in the system.

[0050] Since the present invention is comprised of the aforementioned technical means, when columns with identical data contents are present in different tables, the columns with the identical data contents are handled as the one that belongs to one of the tables (e.g., a table having the largest number of columns to be retrieved) during the process, and all tables that contain identical columns are not always joined. Also, a table that includes no columns to be retrieved is not extracted as tables to be joined. In this manner, an unnecessarily large number of tables can be prevented from being joined.

Problems solved by technology

When desired information is read out from such distributed DBs, it requires very much time and labor to search all servers that manage these DBs for required information.

More specifically, since the user does not know the location of information to be read out in the distributed DBs, he or she must access servers allocated in correspondence with these DBs in turn and must repeat search until he or she finds desired information.

It is impossible to retrieve required information from all the servers unless the user knows the locations (address information such as URL: Uniform Resource Locator) of all DB servers.

However, the distributed DB servers constantly register or delete data, and each DB server itself is constantly connected to or disconnected from the network.

Hence, it is very hard for the user to recognize all these facts and to retrieve accurate information.

However, the search engine can only retrieve the URL information of a DB server, but cannot search an RDBMS (relational DB management system) built in the server at that retrieved URL.

In this way, conventionally, upon acquiring desired data, when DBs that store various kinds of data are distributed, data retrieval requires much time and labor.

Furthermore, in the RDBMS, the maximum number of columns that can be held in one table is normally limited.

However, when such DB having a plurality of real tables A, B, and C is searched for given data, the following problem is posed.

Such processing prolongs the DB search time, and requires a more memory area of the computer that forms the system than required, resulting in low search performance.

However, since the number of columns is also limited on a view provided by an RDBMS as in a real table, a long view cannot be formed beyond the physical limitation.

Images