Forecasting availability of virtual machine based on grey-exponential curve combination model

Abstract

The availability prediction of virtual machine can provide effective guidance for cloud task scheduling and resource allocation. The applicability of existing predictive models is analyzed in this paper, and then a high-precision prediction model is proposed based on grey-exponential curve combination model, which suits the changes in characteristics of availability for virtual machine in cloud. The model is used in the replacement of virtual machine. By predicting and analyzing the availability of virtual machine dynamically, certain virtual machines are replaced. The effectiveness of the prediction model is verified by experiments. The experimental results show that the accuracy of the prediction model given in the paper is better.