| As e-commerce,Internet of things and big data industries are in the ascendant,cloud computing is playing an increasingly important role.In the cloud computing environment,how to reasonably allocate resources for users is an important research content.At the same time,the rapid development of cloud computing has not only promoted the continuous expansion of the number and scale of cloud data centers,but also increased the difficulty of data center management.At present,the problem of low energy consumption and low utilization rate of data center is becoming more and more prominent.Task scheduling technology is an effective method to reasonably allocate resources for users and reduce the energy consumption of data center.On the one hand,it can provide users with computing services and improve user experience by scheduling tasks to appropriate locations;On the other hand,it can schedule non urgent tasks to virtual machines with low energy consumption,so as to reduce the energy consumption of the data center while meeting the user experience.Task scheduling is divided into independent task scheduling and workflow scheduling.This paper mainly studies some problems of workflow scheduling.The specific work is mainly carried out from two aspects: ensuring user quality-of-service(Qo S)and reducing data center energy consumption.The existing work usually only focuses on one of task scheduling,task execution time,user Qo S and data center energy consumption,but few work considers multiple aspects.To solve the above problems,this paper considers the dynamic workflow scheduling to ensure the real-time performance of task scheduling,and takes into account the two objectives of user Qo S constraints and energy consumption,and proposes a new dynamic workflow scheduling algorithm.The specific work is summarized as follows:1.Propose a dynamic workflow scheduling algorithm based on Workflow Partitioning and Improved-Max-Min*.Max-Min* algorithm is improved to make it suitable for dynamic workflow scheduling with time constraints.Using workflow partitioning,the deadline is assigned to the tasks at each level,and the obtained sub deadline is used to calculate the task priority,so that the tasks close to the sub deadline can be scheduled first.Finally,the simulation experiment is carried out on the open source simulation platform Cloudsim.The results show that the algorithm in this paper can effectively reduce the overdue rate of tasks and better meet the Qo S of user quality of service.2.Propose a dynamic workflow scheduling algorithm based on time-cost constraints and energy consumption optimization.By adding cost constraints and energy aware scheduling on the basis of the above work NO.1,different task priority calculation strategy and virtual machine selection strategy are proposed,which can better meet the complex Qo S needs of users and reduce the energy consumption of the data center.3.Propose dynamic workflow scheduling combined with genetic algorithm,and propose primary and secondary strategy scheduling scheme.The main strategy is the content of work NO.2.When the number of ready tasks reaches a certain threshold,some ready tasks are scheduled by sub strategy based on genetic algorithm to get better scheduling results.The experimental results show that this improvement can not only improve the number of successful workflow scheduling to meet user expectations,but also reduce the idle time of virtual machine,so as to reduce energy consumption. |
