Research On Energy Consumption Optimization Scheduling In The Cloud Data Center

With the growth of the cloud computing market,the number and scale of cloud data centers are rapidly expanding globally.In order to maintain their normal operation,the energy consumption problem has become increasingly prominent.High energy consumption not only reduces the return on investment of cloud service providers,but also produces a large amount of greenhouse gases that causes environmental problems,so it is highly concerned by cloud service providers and governments.Eliminating unnecessary energy consumption costs is an important challenge in building the next generation green cloud data center,and it has become one of the research hotspots in the field of cloud computing.The low resource utilization of hosts is an important reason for the energy consumption problem.On the basis of virtualization,reasonable deployment and use of virtual computing units can effectively improve resource utilization,mainly through workflow task scheduling and host resource scheduling.They respectively complete the mapping of the tasks of workflows to the virtual computing units and the deployment and dynamic management of virtual computing units according to their resource requirements.This paper studies the two approaches on two kinds of cloud data centers:VM architecture and two-tier virtualization architecture in which containers are deployed on VMs.The specific works are as follows:(1)For workflow task scheduling,the common deadline-constrained workflow in cloud computing is taken as the scheduling object.The goal is to use the lowest energy consumption to complete the workflow within the deadline given by the user.First,the VM and two-tier virtualized cloud data center models,and the workflow model are constructed.Then,the scheduling process is divided into two phases:time utilization maximization scheduling and running time compression,and TUMS-RTC and TUMS*-RTC*algorithms are given for the two types of cloud data centers respectively.By reducing the number and working time of hosts in the two phases,the goal of reducing the energy consumption can be achieved.Finally,the performance of the two algorithms is tested through simulation experiments using a large number of random workflows with controllable features.TUMS-RTC is superior to the comparison algorithms in terms of VM number reduction rate and energy saving rate.TUMS*RTC*has obvious improvement in resource utilization,VM reduction rate and energy saving rate respectively.In addition,experiments prove that they are suitable for large-scale,highly parallel cloud computing workflows.(2)For the host resource scheduling in VM cloud data center,an energy and Service Level Agreement(SLA)aware VM dynamic consolidation algorithm is proposed,including methods to solve three sub-problems in dynamic consolidation.First,the dynamic independent saturation threshold method is used for overloaded hosts detection,and the concepts of saturation threshold,saturation state and saturation degree are proposed.Second,the underutilized hosts detection uses the combined weight priority method,it uses the adjusted host resource occupancy rate as a ranking index for candidate hosts.Third,minimize number and cost of migrations method is used in the VM selection,it takes the resource occupancy as the selection index.To evaluate the performance of the proposed algorithm,it is simulated in CloudSim and compared with five related algorithms using real-world workload traces.The experimental results show that it outperforms other algorithms,the SLA metric and the energy consumption are reduced obviously.(3)For the host resource scheduling in the two-tier virtualized cloud data center.First,the cloud data center working model that is closer to the actual public container cloud is constructed.Then,an energy-aware host resource management framework is proposed,which includes two algorithms.The initialized static placement algorithm is a two-tier scheduling algorithm,including the load balancing alternate placement and the two-sided matching methods to complete the placement of containers to VMs and VMs to hosts respectively.The runtime dynamic consolidation algorithm includes four methods to realize the dynamic integration of containers.It is dynamically adjusted the initial placement scheme,and utilizes the dynamic consolidation method to use the least active hosts to meet the real-time resource requirements of containers.Finally,simulation experiments are conducted using real workload traces.Compared with related algorithms,two algorithms have better performance in host resource utilization,number of active hosts,number of container migrations and SLA metric.And the entire framework achieves good energy-saving effect.In order to solve the energy consumption problem in the cloud data center,the above work starts from reducing the energy consumption of the host,and proposes effective workflow tasks and host resource scheduling algorithms for the two current mainstream cloud data centers.On the premise of ensuring quality of service,the expected goal has been achieved by improving the utilization of host resources and minimizing the number of active hosts.