 |
 |
|
|
|
To solve the problem of data inconsistency in distributed database, this paper designed a distributed message-oriented middleware to synchronize heterogeneous database. In detail, the paper presents a method of capturing incremental data based on triggers and log tables to get dynamic incremental data. Moreover, the paper presents a message transmission mechanism based on half message and message review to guarantee messages’ order and the exactly-once semantics. In addition, the paper presents a method of generating full data based on incremental data‘s real-time computation to achieve high scalability and high availability. Finally, this paper implemented a prototype system named VENUS. The final test shows that VENUS strictly guarantees the data consistency between the destination database and the source one. Besides, the message transmission of VENUS is the most efficient compared to Kafka and RocketMQ.
|