Energy Saving Strategy With Dual-threshold Rate Adjustment And Its Performance Research

Due to the remain expensive energy costs and increasingly strict environmental standards,high pollution and high energy consumption have become the significant factors restricting the development of cloud data centers(CDCs).Based on the idea of grouping virtual machines(VMs),this paper dynamically adjusts the operating frequency of some VMs according to the system load,and studies the energy saving strategy with dualthreshold rate adjustment(DTRA-ESS)and system performance for CDCs.Firstly,for the CDCs with a low average load,from the perspective of improving energy efficiency and VMs utilization,DTRA-ESS with synchronous shutdown is proposed for non-real-time user requests.By constructing a queuing model with adaptive service rate and partial service desk synchronous closure,the random behavior of cloud users under the proposed strategy is characterized.Numerical experiments and simulation experiments are performed to verify the effectiveness of the proposed energy-saving strategy.Secondly,for the CDCs with a moderate average load,considering cloud users with high real-time requirements,DTRA-ESS with synchronous multiple sleep is proposed.A queuing model with adaptive service rate and partial service desk synchronous multiple vacations is constructed,and performance measures in terms of user request average delay and system energy saving degree are derived.Using system experiments,the effects of parameters such as dormancy parameters and user request arrival rates on system performance are analyzed.Then,for the CDCs with a high average load,the response performance of cloud uses is further improved,and DTRA-ESS with synchronous multiple working sleep is proposed.A queuing model with adaptive service rate and partial service desk synchronous multiple working vacations is constructed,and the expressions of user request average delay and system energy saving level are obtained.Through system experiments,the trend of system performance with parameters such as working dormancy parameters and activation thresholds are revealed.Finally,from the perspective of economics,construct the system utility function by compromising the system energy saving effect and the response performance of cloud users.For DTRA-ESS with synchronous closure,study the Nash equilibrium behavior and social optimal behavior of cloud users,and establish reasonable admission fees to achieve social optimality.For DTRA-ESS with synchronous multiple sleep and working sleep,aiming at maximizing system utility,the optimization schemes of dormancy parameters and working dormancy parameters are given.