Study On Disaster-tolerant System Model Based On Cellular Automata

Along with the extensive application of information systems, companies, enterprises and the government have gradually expanded their businesses to the information platform, which cause the scale of application system being increased unceasingly, the computer nodes involved in operation functions being grown and the coverage area of service application also being expanded. Due to the increasing dependence on the information platform, the high availability and disaster recovery capability of the information system is being a focus concerned in this field.The large-scale service disaster-tolerant system has been in the trend of complexity with the diversity of application, which mainly shows a variety of spatiotemporal complex. Besides, its status and responsive events are discrete; and there is a characteristic that the operation of the network topology, the service node, the duplication and the detection are in dynamic combination. While it's difficult for the traditional technique to accommodate the need of modern disaster-tolerant system research, the cellular automata, as an efficacious tool to study system discrete in time and space, can be used for analyzing and estimating the large-scale disaster-tolerant system under the complex application environment.As one important tache of data reproduction, the asynchronous duplication transmission is the foundation for the whole disaster-tolerant system to operate normally. In the foundation of analyzing the existed mechanism of the asynchronous duplication transmission and its limitation, one improved mode that be auto-adjusted duplication strategy with consistent of network condition is proposed. The simulation analysis and the research to the above pattern model has been carried on by using the cellular automata, in which it can be seen that under the mechanism of network asynchronous transmission there being spatiotemporal long-range dependece nearby the congestion of the critical point. So according to the above properties, this improved duplication pattern may act a certain extent enhance on transmission performance.With the combination of above analysis of the asynchronous duplication transmission mechanism, one failure-detector model based on the certainty factor and the voting mechanism has been proposed. Once again, with the simulation analysis and the research to the improved model by using the cellular automata, it is clear that the indices such as model accuracy and detection time are all conformed to the required performance of modern disaster-tolerant system.As to the issue that there is still a lack of effective evaluation method of performance analysis, a new evaluating method has been put forward in order to estimate the disaster-tolerant capability and the cost of the specific complex service system. Through modeling and analyzing one case based on the peer-to-peer storage systems, the optimal point of the disaster-tolerant construction to the certain case can be found, which provides the theory reference and the gist for constructing reality disaster-tolerant system.