Research On Cloud Computing Task Scheduling Strategy Based On Ant Colony Algorithm Under QoS Constraints

Task scheduling is an important part of cloud computing operation process,which is related to the overall performance of the scheduling system.The three level service system of infrastructure,platform and software based on cloud computing is related to task scheduling and resource allocation.A good task scheduling algorithm can shorten the time of task completion and reduce the cost of task completion,so as to improve user satisfaction.Based on the dynamic and scalable feature of cloud computing,it is a hot topic in task scheduling to meet the change of user needs.Ant colony algorithm is widely applied in cloud computing task scheduling to solve NP-hard problem,and achieves better scheduling results.But the defect of ant colony algorithm is the lack of initial pheromone and easy to fall into the local optimal solution.The existing scheduling algorithm's optimization goal is relatively single,and most of the scheduling models only consider one of the targets including task completion time,load balance,user satisfaction and so on.For user tasks with deadline completion time and expected cost QoS(Quality of Service),the scheduling strategy on optimizing user satisfaction is considered less.This paper proposes a task scheduling strategy based on ant colony algorithm under QoS constraints in cloud computing.The research in this paper is mainly the mapping relationship from multi-dimensional tasks to multi-dimensional virtual machines.Making full use of virtual machine resources,we can reduce consumption time and improve user satisfaction.Considering the user's deadline of completion time and the expected cost of tasks are under two QoS constraints,time and cost membership functions are established respectively.The normalization of two evaluation indexes is defined as the user satisfaction function,and it is integrated into the global update method of the pheromone of the ant colony algorithm,which dynamically regulates the scheduling of the task.When the pheromone is initializing,it takes account of the length of task and the execution ability of virtual machine,and it improves the defect of lacking pheromone and accelerates the convergence speed.The virtual function is added to the heuristic function to adjust the load balancing of the virtual machine.In order to improve the overall performance of the algorithm,the parameter selection of the algorithm is studied and the range of its value is determined.The results of CloudSim simulation experiments show that the task scheduling strategy based on ant colony algorithm under QoS constraints can reduce the task completion time and improve the user satisfaction.