Reliable Design And Realization Of The Distributed System Oriented To Redundancy

As the development of the computer technology, the application of the computer becomes more and more widely, and human's daily life more and more depends on the computer's reliability. If the computer system fails, great damage would be caused. The distributed system is widely used for its outstanding advantages such as robust, easy scalable, quick response and shared resource. So how to improve the reliability of the distributed system becomes a research focus.On research of the distributed object tree-management model that oriented to redundancy, the cause of the single failure problem that often happens on the distributed system was analysed. A reliable distributed system management model is given in this paper. This reliable model is realized through several processes which include error-detection, error-exclusion, and object switch. CORBA and Java technology is used in realizing the reliable model. The reliable model can detect the object failure and get rid of the error automatically.The conception of object redundancy is brought forward creatively in this paper. An two module redundant object technology is used in the reliable model. One of the pair of the redundant object acts as a main object, the other acts as a slavery. An hiberarchy error detection method is used by integrating The check-point detection technology and the heart-beaten detection technology, this method can decrease the communication cost and keep the detection veracity at the same time.