Research On Cloud Virtual Machine Energy-saving Strategy With Semi-sleep Mode And Performance Of Strategy

With the increasing number of cloud resources and the types of cloud computing services,the energy consumption of cloud data service center increases.While ensuring the energy efficiency of the cloud data service center,better meeting the response requirements of cloud user request,introduces semi-sleep mode and awakening threshold,and studies the cloud virtual machine(VM)energy-saving strategy and the performance of strategy for the cloud data service center.Firstly,from improving the response performance of cloud users point of view,taking into account the energy-saving rate of cloud data service centers,a cloud virtual machine(VM)energy-saving strategy with semi-sleep mode is proposed.When there is no cloud user request in the system,the virtual machine(VM)enters the semi-sleep period,and the virtual machine(VM)during the semi-sleep period can provide low-speed service for the randomly arrived cloud user request,which can ensure the system energy-saving rate and improve the response performance of the cloud user request.Secondly,from the further enhancement of the energy-saving effect of the cloud data service center,and prolong the time for the virtual machine being in a semi-sleep state,a cloud virtual machine energy-saving strategy combining the semi-sleep mode and the awakening threshold is proposed.When the semi-sleep period ends,if the number of cloud user requests waiting in the buffer has not reach the awakening threshold,the virtual machine will independently start a new semi-sleep period and continue to provide low-speed service for cloud users request to achieve the purpose of saving energy.Then,for the cloud virtual machine energy-saving strategy with the semi-sleep mode,a multi-server system model with asynchronous multiple working vacation is constructed.For the cloud virtual machine energy-saving strategy combining semi-sleep mode and awakening threshold,a multi-server system model with asynchronous multiple working vacation and N-policy is constructed.Using the quasi-birth-and-death process(QBD)and the matrix geometric solution method,the performance indicators,such as the average response time of cloud user request and the system energy-saving rate,are derived.Finally,as a result of theoretical analysis and simulation experiments,the effects of semi-sleep parameter and awakening threshold on the performance of cloud virtual machine energy-saving strategies are carried out.A cost function of system is established based on the trade-off between the average response time of the cloud user request and the system energy-saving rate.The ant colony intelligent optimization algorithm is improved to achieve the goal of minimizing the cost function of the system,and the system parameters are optimized for the cloud virtual machine energy-saving strategy.