Research Of Data Replication In Distributed Database System

Data replication is a very important technology in the distributed database, which can improve data availability and system performance. Owing to it's accessing local data and the avoidance to access remote data using redundancy, it can not only reduce communication cost and response time but also run system rightly even in presence of sites and communication line failure. This thesis mainly analyses the basic theories of distributed database system and data replication technology, discusses detailedly the key technology that the system realizes, proposes the architecture and algorithm of data replication, and summarizes the implementation schema of heterogeneous database replication.The thesis proposes a new data replication scheme based on Loose Consistency Agent.The framework defines two types of nodes: master node and slave nodes; then defines four components: capturing manager, distributor, synchronizer and conflicts manager. Capturing manager uses a capturing method based on change trace table, distributor uses a copy partition algorithm, synchronizer broadcasts update, conflicts manager can detect and resolve conflicts through vector clock. In addition, data transmission algorithm and clock synchronization algorithm are used between master node and slave nodes. In a word, the scheme can solve the bidirectional data replication over databases.According to the replication scheme based on Loose Consistency Agent and other theories, a prototype tool named Unaided Replicator is developed. The testing results show that Unaided Replicator actualizes the characters of above theories, and achieves our expectation.