| As an emerging service model of business computing,cloud has brought a new round of innovation in both industry and academia.The elasticity and convenience of cloud contributes to its ability of processing massive data simultaneously.Considering the poor resource utilization in datacenter and the lengthy waiting time of users,it can be found that the current optimization goal of cloud computing is relatively single,considerations cannot be given to both workflow time overhead and datacenter resource utilization.That is why the study of mapping the workflows to resources in a time-ware and efficient manner is of great importance.Based on the analysis of relational researches in the domestic and abroad,considering the diversity of workflow application and the fact that cloud users pay more attention to the time makespan while cloud providers is far more concerned about resource utilization,a workflow time makespan model for users and a resource utilization model for cloud datacenter are presented in this dissertation,and the following work contents are taken:(1)In view of the above challenge,a dynamic scheduling method for concurrent workflows,named as DSM,is put forward.The strategy is designed based on critical path lookup,time and utilization aware task scheduling and priority driven resource preemption,minimizing the time makespan of workflows,and maximizing the resource utilization of datacenter during the execution of the workflows.The advantage and feasibility of the proposed strategy are verified through the analysis of simulation results.(2)With revised workflow time overhead model,a workflow task scheduling method based on genetic algorithm is proposed,in which the chromosome selection,crossover and mutation operation are modified by combining with task priority and real-time resource status to improve the convergence rate.Experimental results show that the presented algorithm has superiority performance in optimizing time overhead and resource utilization. |
PDF Full Text Request