Research And Implementation Of High Availability Technology For Distribution Massive Database System

With rapid development of information society, a huge amount of data were produced and stored in database, the volume of database storage has reached to a magnitude scale. Distributed massive databases system which stores data in multiple databases is widely applicable to many fields. Because of the huge scale, numerous components and complicated relations, availability of a massive databases system has become a serious challenge. Many research of availability for database system locates to some layer such as physical layer or application layer. But it is not suitable for Distributed massive databases system.Based on analyzing a distributed massive database system which is running some key applications, we study some high availability techniques and propose a Multi-layer architecture for distributed massive databases system, and then we use our design to the real system. By observing the running of the real system with our high availability techniques we found our design obtained good results.In our multi-layer architecture for distributed massive databases system, backup DB Group can detect DB fault quickly and avoid server not available for single database error. On middleware layer, fault tolerant server implements multi-replica server dynamic switch by redundant server replica. We also design backup fault tolerant server to avoid simple point failure, and we also propose a plan to locate server based on the availability model.By testing and observing the running of the real system, our high availability techniques improve the availability of the system.