Research On Workflow Tasks Scheduling Strategy In Cloud Computing

In recent years,cloud workflow tasks scheduling has been an important research topic in the business community.Cloud workflow tasks scheduling refers to the fact that the workflow tasks submitted by the user are allocated to the appropriate computing resources and the corresponding costs are paid in real time according to the usage of the resources.For most ordinary users,they focus on two quality of service indicators: the completion time and execution cost of the workflow tasks.Therefore,how does the cloud service provider design a scheduling algorithm to optimize the completion time and cost of task is the issue of concern.Based on the idea of workflow,the current research aimed at studying the workflow tasks in the cloud computing.I mainly studied contents as follows:(1)Aiming at the limitation of HEFT(Heterogeneous Earliest Finish Time)algorithm to optimize workflow task completion time.The idea of task migration was introduced,and a single objective optimal scheduling algorithm based on task migration was proposed.The algorithm achieved mission-critical and non-critical tasks after the task was pre-scheduled by HEFT algorithm,then migrated non-mission tasks to low-cost virtual machines while ensuring that the task's completion time was constant.Finally,the simulation experiment was carried out by CloudSim,experimental results show that the scheduling algorithm based on task migration was the same as the task completion time of HEFT algorithm,but it was less expensive than HEFT algorithm.(2)Aiming at the problem of dual-objective optimization of cloud workflow execution cost and completion time,a multi-objective optimization scheduling algorithm based on task replication was proposed.The algorithm made full use of the free time of the virtual machine while the task was scheduled and reasonably replicated the predecessor task of the current task so that the task can be completed ahead of time and reduce the execution cost.Finally,through the removal of redundant tasks to further optimize the cost.Simulation experiments show that the proposed algorithm based on task replication not only can optimize the task completion time,but also effectively reduce the total execution cost of the task.