Research On Intrusion Tolerance Base On Byzantine Algorithm Under Cloud Computing

Cloud computing is a promising technology to facilitate the development of large-scale, on-demand, flexible computing infrastructures. The basis of cloud computing is virtualizationTechnology, many computing tasks are executed in a virtual resource pool which is constituted byphysical computing resources. according to user requirements, the system automatic allocates thecomputing or storage resources that user need. However, the cloud computing platform still havemany problems, one of the problems is that cloud computing platform data intrusion tolerantperformance is still to be improved.The existing technology in a distributed system research intrusion tolerance technology focuson signature code detection and distributed redundancy replication. These technologies requirehigh-performance system hardware and software, and in the existing intrusion tolerance technology,the recourses in system have large overhead and low utilization rate, The high cost of intrusiontolerance is unacceptable for cloud computing service providers and users, it is difficult to satisfy theneeds of cloud computing.In order to solve the intrusion tolerance of cloud computing platform much better, firstly, thispaper studies the Byzantine algorithm deeply, and analysis its working principle and algorithm.Secondly, a comparative analysis is conducted on the flow of work and the performance of thealgorithm of two kinds Byzantine algorithm——MinBFT and the MinZyzzyva, this paper proposesa new algorithm that the USIG services in MinBFT algorithm is introduced into MinZyzzyva, by theuse of the USIG service, MinZyzzyva algorithm is able to receive requests in order of priority, whenthere are a large number of requests which have different priority, system can improve the efficiencyand performance of the algorithm. Thirdly, the improved MinZyzzyva algorithm can tolerate F errorreplicas in2F+1replicas, the improved algorithm reduces reduce resources consumption on cloudcomputing, thereby improve the intrusion performance of the system.Lastly, on the Hadoop open-source platform, using the java programming language to achievethe improved MinZyzzyva algorithm, this paper contrasts the existing PBFT, MinBFT andimproved MinZyzzyva algorithm in three aspects including throughput, delay and number ofreplicas. The experimental results show that MinZyzzyva algorithm has a better performance inintrusion tolerance and reduces the cost of cloud computing greatly through improved MinZyzzyvaalgorithm by introduce the USIG services in MinBFT algorithm into it.