Multi-resource Power Aware Scheduling Oriented Simulation Tool,Algorithm And Research For Cloud Computing

Cloud computing is rapidly growing and widely used thanks to its high availability,scalability,and on-demand services,but the increasing growth in energy costs of cloud data centers continue brings environmental issues that are one of the major facing challenges.Energy issues involve multiple subproblems: virtual machine scheduling,multi-resource scheduling,job scheduling,load balancing,energy awareness and so on.The emergence of virtualization technology and virtual machine scheduling enhances the resource utilization of the server and reduces the number of hosts used thereby saving energy.Similarly,a reasonable scientific job scheduling strategy can improve resource utilization,reduce job completion time and energy consumption.In addition,the reality of the cloud environment,the evaluation of a candidate algorithm is often very difficult,because the user needs are complex and diverse,the task load changes constantly,the fair environment is difficult to reproduce,the expriment costs time and consumes resources,testing needs repetition and parameters need training.Therefore,cloud simulation tools become very meaningful.As a leader in cloud simulators,it can be used easily on the virtual machine scheduling problem and job scheduling problem.But CloudSim is not so perfect,in the multi-resource scheduling and energy consumption simulation,CloudSim there are still some deficiencies.In view of the above problems,on the basis of understanding the domestic and foreign research results,this paper makes the following work on these aspects:(1)A virtual machine joint placement algorithm based on peak workload characteristic is proposed.The complementarity of virtual machine historical load is described by peak workload similarity,and the virtual machines with high complementarity are placed on a host.The theory and experiment show that the algorithm can effectively improve the utilization of the hosts and reduce the number of hosts.(2)Considering the shortcomings of CloudSim in multi-resource scheduling and energy consumption simulation,we developed the novel simulation tool called MultiRECloudSim,which implements multi-resource scheduling,job assignment,multi-resource allocation and resource allocation mode,multi-resource energy consumption simulation and other functions.Experiments show that its function is rich and convenient.(3)The prime resource load balancing assignment algorithm in gomogenous / heterogeneous environment and the time balancing assignment algorithm are proposed.The experiment of MultiRECloudSim proves that the prime resource load balancing assignment algorithm can effectively reduce the energy consumption of data center,the time balancing algorithm can effectively reduce the job completion time.