Low-energy-consumption Workflow Scheduling Method Based On Combined Allocation In Cloud Environment

Workflow abstracts,summarizes and describes the execution rules between business processes and related operating steps,and its technology has been widely used in science,industry,and engineering.With the continuous development of cloud computing technology,more workflows are deployed under the cloud environment with computing services which can be employed on demand.As is known to all,energy consumption is a major problem facing cloud workflow scheduling.Currently,voltage/frequency dynamic adjustment technology(DVFS)or shutting down low-utilization servers are two common methods to reduce energy consumption of cloud servers.However,they often inevitably lead to an increase in the execution time of workflow tasks while reducing energy consumption,which leads to a dilemma.Therefore,existing scheduling methods often rarely strike an effective balance between reducing execution energy consumption and shortening completion time.To this end,this thesis proposes a workflow low-energy scheduling method based on merged allocation to divide workflow scheduling into three stages: task merging based on the longest path,server scheduling based on merged task strings,and task slack based on scheduling results.Task merging based on the longest path repeatedly finds the longest path in the workflow and merges the tasks that meet the conditions in the path to reduce the completion time of the workflow as much as possible.The server scheduling based on the merged task string judges the tasks that have not yet participated in the merge in the first stage on the basis of the first stage to complete the further merge.After that,the generated task strings are used as the basic scheduling unit for scheduling,and each task string allocated to as few servers as possible to initially reduce the energy consumption of workflow execution.Task slack based on scheduling results on the basis of the scheduling results completed in the second stage,the available time slots of each server and the slack time of each task are obtained,and the relevant time slots are comprehensively slacked based on DVFS technology,so as not to extend the completion time Under the premise of further reducing energy consumption.Finally,specific examples and simulation experiments verify the feasibility and effectiveness of the scheduling method in this thesis.For the two types of input of some classic scientific workflows and randomly generated workflows,the comparative data shows that this method can shorten the completion time of the workflow and reduce execution energy consumption effectively at the same time.