Research Of Virtual Resource Utilization Rate Prediction In Cloud Environment

Since cloud computing technology is introduced,it has made great progress and attracted many companies and individuals' attention.Now it has been used in many industries.Cloud computing technology has powerful storage capacity,computing power and scalability.Nowadays,people can purchase cloud computing services according to their own needs,which saves software and hardware costs for many individuals and small enterprises.However,in practical application,with the increasing number of users,the scale of the cloud platform is also increasing.There are hundreds or even more virtual machines in the virtual resource pool.The application system of each virtual machine occupies different resources and is constantly changing at different time.These states are difficult to grasp.For the use of virtual machines,there is no platform for accurate and long-term statistical analysis.Operations and maintenance personnel cannot accurately grasp the use of resources and demand trends,which brings great inconvenience to the management of cloud computing platform.To this end,a virtual resource utilization analysis system was designed to track,count and predict the utilization of major virtual resources such as CPU and memory.Firstly,the overall framework of the system is designed,and the functions of each module of the system are described in detail.Secondly,BP neural network and genetic algorithm(GA)are deeply studied.Using GA to optimize BP neural network overcomes the shortcoming that BP network falls into local minimum value easily.Then,the Hadoop platform and the web platform are built,and the idea of parallel programming of GA and BP neural network based on MapReduce are given.Finally,the system is developed by using javaee technology.The results show that the system can track the running status,CPU utilization and memory utilization of each virtual machine on the cloud platform in real time,and also can calculate the maximum utilization,minimum utilization and average utilization of CPU and memory in a certain period of time.The system could export resource reports.With the parallelized GA-BP model,the CPU utilization and memory utilization requirements can be predicted quickly and accurately.