Energy And Reliability-Aware Cost Optimization Scheduling For Cloud Workflow Applications

With the continuous increase of data scale and computational complexity,executing modern workflow applications in cloud computing environments will involve a large number of cloud resources of different types and prices.This makes the cost of cloud workflow scheduling a focus of attention.On the other hand,as the energy consumption of cloud data centers is also increasing,the energy consumption of cloud workflow scheduling has also become a concern of academia and industry.In order to provide users with lower-cost workflow scheduling services while reducing the energy consumption of cloud data centers,performance-based pricing schemes came into being.The cloud service provider can dynamically adjust the processor voltage/frequency based on the dynamic voltage and frequency scaling(DVFS)technology to reduce energy consumption,and adjust the price according to the changing frequency of the virtual machine(VM)when it is running.However,improperly lowering the voltage/frequency will increase the probability of soft errors and reduce the reliability of the workflow.Therefore,how to provide users with the cost-effective cloud resource provisioning and workflow scheduling solutions,while considering energy consumption and reliability requirements is a major challenge for researchers.In response to the above problems,this paper proposes an energy and reliabilityaware workflow cost optimization scheduling method,which can effectively generate energy-saving,reliable,and near-cost optimal cloud resource provisioning and workflow scheduling solutions for cloud workflow applications.The main contributions of this paper are as follows:1.This paper analyzes the impact of DVFS,soft error,and checkpoint fault tolerance mechanism on cloud workflow scheduling in detail,and accurately models the cloud workflow scheduling system.The cost optimization problem of cloud workflow scheduling under makespan,energy consumption,reliability,and memory constraints is defined formally based on the cloud workflow scheduling system model.2.This paper proposes an energy and reliability-aware cost optimization workflow scheduling algorithm.The algorithm can reasonably provide cloud resources,allocate tasks to VMs and dynamically adjust the operating frequency of the VMs,thereby generating a near cost-optimal scheduling scheme for the workflow while satisfying makespan,energy consumption,reliability,and memory constraints.3.This paper implements an energy and reliability-aware cloud workflow scheduling tool,which implements the energy and reliability-aware workflow scheduling system model and the algorithm.The tool can simulate the cloud data center and scheduler that support DVFS and performance-based pricing,simulate workflow scheduling under the influence of soft errors and checkpoints,and generate and evaluate the workflow scheduling solutions.Comprehensive experiments on various well-known scientific workflows validate the effectiveness of the proposed method.Comparing with the three state-of-the-art scheduling methods based on DVFS,the proposed method can significantly reduce the overall cost and energy consumption without violating the given constraints.