DATA PARTITIONING METHOD FOR DISTRIBUTED PARALLEL DATABASE SYSTEM

摘要

A data partitioning method for distributed parallel database system includes the following steps: a fact table and a dimension table are established according to the built distributed parallel database system; the records of the dimension table and the fact table are inserted into different nodes according to the partitioning rule; the record of the dimension table is copied into the node of the fact table; and the data deletion and renewing processes are implemented. When partitioning inputting or inserting data set or data flow into the distributed database system, the present invention can satisfy, in each node, the relation between the tables which is defined in the database project, especially the Primary-Foreign Key restriction condition, and provide the data of each node with local maturity. For the query of the connection between tables utilizing the Primary-Foreign Key restriction condition, because the data of each node is provided with local maturity for the type of query so that no dynamic data re-partitioning is need between the nodes, the time consumption for data network transition is avoided, the query response time is reduced, and the query efficiency is increased.