Cloud Data Centers Energy-saving Scheduling Algorithm Based On CPU Frequency Scaling

High energy consumption of cloud data center has become a serious problem.In recent years,the increasingly complex organizational structure and increasing scale of data center make energy consumption management more challenging.The main works of this paper includes:building of OpenStack Essex cloud management platform;creation of physical machines resources monitoring system;creation of energy consumption model of CPU frequency scaling;design and implementing of virtual machine scheduling algorithm in OpenStack system;several groups contrast experiments and experiments results analysis.This paper lays stress on the different performances between SCFS and other traditional scheduling algorithms when virtual machines loading different kinds of applications.SCFS is based on CPU frequencies scaling technology,in virtual machine initial placement stage it makes out a multi-objective solution by heuristic ant colony algorithm to decrease energy consumption and better meet SLA;In virtual machine dynamic management stage it decrease virtual machine migrations used in traditional method by auto-regressive prediction and CPU frequency scaling technology to decrease migration cost.The experimental results in the cluster of21Lenovo servers shows that the scheduling algorithm based on CPU frequency scaling perform much better than the traditional Best Fit Descending and First Fit Descending algorithm in energy saving and SLA meeting.This paper analyses some shortcomings existing in experiments and proposes improving measures and future researches at last.