Research On Task Scheduling Of QoS Target Optimization In Cloud Computing

With the popularity of the Internet and the rapid development of information technology,the cloud computing industry is also developing rapidly.Its "on-demand services,according to the amount of payment" has become a typical feature of cloud computing.However,cloud computing systems require different QoS.Different requirements,computing node resources to cloud computing task scheduling problem becomes very complex.Task scheduling is a key research problem in cloud computing.A scheduling strategy directly determines the degree of user satisfaction,task completion time,system resource utilization rate.In order to solve the problem of cloud computing task scheduling,there are many scheduling models.The traditional task scheduling model is more concerned about the efficiency of load balancing,optimal span etc.They consider less about the completion of the task and the task requirements to complete the cost of QoS requirements.The cloud computing is a business model.It's important to protect the user QoS requirements.Therefore,this thesis starts from the intelligent optimization algorithm and the scheduling model,and considers the QoS requirement and task completion time of the user to complete the task completion time and the task completion cost to reduce the task completion time and improve the customer satisfaction.This thesis first studies the task scheduling model in cloud computing and the existing problems in the existing task scheduling.Second,combined with the user task scheduling QoS requirements-task deadline completion time and task expectations to complete the cost,respectively,the establishment of membership function to quantify the user needs.Finally,the objective function is defined and constrained according to the scheduling goal.In order to achieve the scheduling goal,an improved differential evolution task scheduling algorithm based on QoS target optimization is proposed.To reduce the task completion time and improve the efficiency of the algorithm,the algorithm initializes the 2 * NPindividuals and introduces the fast sorting algorithm to filter the NP as the initial population to obtain a higher fitness value for the individual.In the process of variation,the mutation strategy EbestD1// is introduced on the basis of the standard variation strategy,and the two strategies are integrated into the algorithm to accelerate the convergence of the algorithm.To facilitate the effectiveness of the proposed scheduling policy evaluation,compositions herein further study scheduling algorithm parameter selection strategy,and suggests selected value.Finally,the simulation experiment is carried out on the Cloud Sim experimental platform.The simulation results show that the scheduling strategy can improve the convergence speed,reduce the task completion time and improve the user satisfaction.