Research And Implementation Of Mediator System

Interested data from different data sources such as Web, OLTP (online transaction processing) database, can be gathered, well organized and integrated quickly and accurately into virtual database, which is a new database technology, and afforded to decision maker and application softwares in a uniform relation database form. Therefore, the integrated data lay a basis for data mining and OLAP (on-line analysis processing) and good decision support can be provided. Based on the mediation technology, this paper is completing and improvinging a cross-OS and cross-DBMSmediator database system which aims at structural information sources that are supplied by relational databases.First, in the paper, according to the actual demands we revise the system model and each System functional modules. In these modules, Compiler for SQL transforms SQL queries to relational algebra expressions and supplies a mechanism for disposing errors. Management system for metadata creates hypergraphes to build the foundation for query translating and extract and anatyze metadata of information sources and finishes the metadatabase management. Query translator creates automatically join sequences and tests lossless joins, then decomposites and composites the queries, so that the mediator database can accomplish the lossless translation that translates the queries from the mediator database to the information sources. Query broker implements universal database management, query rewriting, query file generating, maintenance of concurrent queries and query result files so as to complete the request of queries from the mediator database to each information source. JDBC driver implements the JDBC standard API for the application. Finally, the paper also gives the system the specific method and the code structure.