Research And Development The Key Issues Of High Availability Database Middleware

With the developing of the Internet technology, database technology has become the lifeline of the national economy. It is the foundation of the electronic commerce, the E-government,MIS,ERP,as well as financial system. So it is very important to increase its availability. However, with the data quantity increase and the high concurrent visit's appearance, cause traditional the stand-alone database unbearable, so the system's availability urgently need to be solved. The cluster technology is the first choice, the deep research database cluster technology will be helpful in enhancing its performance and availability, and the high availability as an important measure of database system, this paper will focus on it.Based on the analysis of the current availability technology of the database, we design a common database cluster system model. It is using a high cost-effective database middleware to achieve parallel processing. We design and implement the database cluster middleware C-RAIDb. The middleware is used for the database cluster and fault-tolerant service. It can hide the complexity of distributed database, and gives the client a single system image. C-RAIDb system uses a series - parallel architecture, the system application layer, middle layer and resource layer composed of a serial. Because the availability of serial system does not exceed the availability of the lowest layer, in order to increase the availability of the middle layer, the middle layer provides a number of virtual database parallel processing the request. It is use JGroup to synchronize virtual database from different controllers. The horizontal expansion of C-RAIDb controller can increase the capacity of the controller's fault-tolerance. In this structure, a node is not only served as a controller but also as a database server; vertical expansion of C-RAIDb controller can be applied to the case of a large number of database background. At this time, the controller and the database server are different nodes. In resource layer, we use database cluster to achieve the data high availability. For the data's replication and distribution, we introduce the concept of RAIDb, replicating the data between nodes with the form as basic unit. In order to perform better fault tolerance and performance, we combine the database instance to a database array. Finally, we use the middleware to build up a database cluster. And we test its data consistency and single point of failure in the database cluster system. The results show that the database cluster based on the middleware has a good availability.