Research On Method Of Query Decomposition And Concurrency Control In Heterogeneous Database System

Heterogeneous database system (HDBS) may have been shielding the different access methods and user interface of a various of heterogeneous data management systems. It provides a public interface for user to visit a variety of heterogeneous data sources. At the same time, it provides an information processing platform which integrates multi-data sources and merges a number of data query results. Query processing is one of the key technologies in heterogeneous database system. Query decomposition and query concurrent scheduling are the core contents for query processing.Aimed at the differences of data mode and semantics in global database and local databases, this paper analyses the characteristics of the heterogeneous database system, and redesigns the data dictionary of the heterogeneous database integrated system. The data dictionary includes the global data dictionary, the local data dictionary and mapping from global mode to local mode. In addition, the data dictionary increases the table of intermediate results status information, which used to store the information of intermediate results during the process of global query decomposition. It is more convenient to merge into the final results that according to the information of intermediate results.Based on the differences of data modes of each local data sources in heterogeneous database system, it proposes RAC notation of SQL language. The RAC notation gives an agreed indication of SQL language; it only involves database relation R, attribute A and query condition C in SQL language, and shields the differences of query sentences in each data sources.It designs decomposition algorithm of global query which based on RAC notation of SQL language. Thus, it realizes decomposition from global query to local sub-query accepted by each local data sources, and completes transform from global mode to local mode. It needs to send the decomposition results to the corresponding local data sources for query or update operations.This paper analyses and researches the problem of concurrency control in query processing between heterogeneous databases. Combined with 2PL protocol and serialization precedence diagram, it proposes a concurrent scheduling technology based on Operation-Lock in heterogeneous database system. The concurrent scheduling technology locks the whole tuples which involved in basic query or update operation, so that it increases the concurrent degree of transactions to a certain extent. Finally, the paper gives the proof of accuracy and effectivity of the concurrent technology.