Cloud Computing Resource Scheduling Strategy Based On ARMA Model Prediction

At the current stage, cloud computing is favored by more and more enterprises as a new type of efficient and low-cost computing model and the study on cloud resource scheduling algorithm which is the core research of cloud computing has never stopped. Generous research has been carried out on the cloud resource scheduling algorithms. This paper introduces some of the current mainstream cloud resource scheduling strategy and analyzes the advantages and disadvantages of these strategies first. Considering that problems of the host switch machine fluctuation, the blindness of the virtual machine migration and the long task response time still exist in current cloud computing resource scheduling strategy, we presents a cloud computing resource scheduling strategy based on time series prediction. The strategy includes two modules, namely, the time series prediction module and the cloud resource scheduling control module. Among them, the time series prediction module mainly establish time series forecasting model based on the historical load data of cloud computing data center, use the time series model to make real time prediction of load variation trend in the future on cloud computing data center or designated host machine. The cloud resource scheduling control module realize pre-allocation of resource and optimization of the host machine switch process and virtual machine migration process based on the cloud resource load prediction value provided by cloud resource load prediction module. Through the resource variation analysis between actual load value of cloud resources at present time and cloud resources load forecast value at the next moment, we can infer the relationship between supply and demand of resources from the current moment to the next moment to achieve the dynamic adjustment of data center resources load of cloud computing. Finally, we make the block-based implementation about cloud computing resource scheduling algorithm based on time series prediction in open source cloud computing simulation platform Cloudsim and mathematic mathematics software Matlab. Concrete implementation procedures are as follows: We realize cloud resource load prediction module in mathematic mathematics software Matlab and cloud resource scheduling control module in open source cloud computing simulation platform Cloudsim, and finally we achieve the integration of the two modules in the algorithm by mixed programming approach. Through the simulation experiment of multiple original resource scheduling algorithm and improved algorithm in this paper separately and the analysis of experimental results, we can verify validity and generality of cloud computing resource scheduling strategy based on time series prediction in this paper.