Research On Workflow Scheduling Algorithm Under Cost Budget In Heterogeneous Cloud System

It is difficult for traditional high-performance computing to satisfy the diverse demands of computing resources,cloud computing.As a pay-asyou-go service model,cloud computing has been developing rapidly.Cloud computing provides users with various resources and effective computing platform for the deployment of large-scale workflow applications.Cloud computing brings convenience to users around the world under the era of green computing.However,large-scale data computing accompanied by huge energy consumption has caused irreversible harm to the ecological environment.With the gradual maturity of public cloud pricing models,which brings new challenges to the workflow scheduling problem under budget constraints in heterogeneous cloud computing systems.Based on the above background,this thesis focus on optimization problems of makespan and energy consumption for workflow under budget constraints in heterogeneous cloud environment.The main innovations and research work of this thesis are summarized as follows:(1)A cost budget constraints and efficient workflow scheduling algorithm(ESBL)in heterogeneous cloud computing systems is developed while considering the problem of minimizing workflow execution time.Firstly,the task priority and budget-level cost of each task are established according to the given set of workflow tasks and budget cost.Then,the critical virtual machine nodes corresponding to the critical path in the task set are determined in according with the given virtual machine clusters.Finally,tasks in the critical path of workflow are assigned preferentially to the critical virtual machine,and the other tasks are assigned to the best virtual machine according to their updated budget cost.(2)Energy consumption is an important factor that affects cloud system performance.A reducing energy consumption strategy using a critical task remapping(RMREC)algorithm is proposed in this thesis.Firstly,the adjust cost budget and spare cost are determined on the basis of cost budget,critical task path,and adjustable budget factor.Then,the best combination of virtual machine and frequency is selected according to the updated budget level cost of each task to realize the preliminary mapping between tasks and virtual machines.Finally,critical tasks are remapped to VMs according to spare cost to decrease energy consumption caused by task migration.(3)Simple heuristic algorithms cannot meet the actual workflow task scheduling requirement well.A workflow scheduling strategy based on the whale optimization algorithm is devised to minimize energy consumption.In the beginning,the initial population is generated by opposition-based learning.Then,the position of tasks is updated according to the size of the random number.Lastly,the worst individual is replaced by the group optimal position in order to avoid falling into the local optimum in the iterative updating process.The extensively experimental comparison results show that the presented algorithms can effectively reduce the scheduling length and reduce data center energy consumption while meeting the budget constraints.