Research On Task Scheduling-based Intelligent Optimization Algorithms In Cloud Computing

Cloud computing is the latest distributed computing paradigm and it offers tremendous opportunities to solve large-scale complex task scheduling problem.However,due to the complexity of cloud environment,the heterogeneity of resources and the high charging of cloud services,it faces great challenges in providing reliable and high-quality services.Task scheduling plays an important role in reducing task completion time and user costs,therefore,it is a hot topic that efficiently designing intelligent algorithms to schedule cloud tasks and improve the quality of cloud services in current cloud computing research.In this paper,we propose intelligent optimization algorithms for independent task scheduling,workflow scheduling and priority task scheduling based on their characteristics,respectively.The main works are as follows:1.In terms of the independent task scheduling,due to the features and complex scenarios of the cloud,traditional scheduling approaches face with three challenges: robust model,local optima and slow convergence.First,we established a robust task scheduling model.Then,a particle swarm optimization method based on Self-Learning Strategy and Nearest Neighbor Heuristic mechanism(SLNPSO)is proposed.The self-learning strategy is used to improve the diversity of population and the neighbor heuristic mechanism can accelerate the convergence speed.In addition,a greedy policy was introduced to quickly initialize particles,so as to improve the quality of the initial solutions.We conducted simulations on CloudSim platform,the proposed algorithm has a fast speed of convergence and can avoid trapping into the local optimum.Meanwhile,the algorithm can reduce the time and cost overhead in the process of task scheduling.2.For the workflow scheduling in cloud,aiming at the impact of complex workflow structure on scheduling performance and the increase of cloud services in cloud computing,we propose an efficient task-clustering based cost-effective aware scheduling algorithm(ECOS).In order to minimize the cost within the deadline of workflow,we have devised ECOS with two key steps: vertical clustering based on the time consideration that selectively merges the sequential tasks to reduce the transferring time of workflow;horizontal clustering and greedy allocation are used to aggregate the parallel tasks,and greedily allocate resources to tasks with the aim of minimizing cost within deadline.We have conducted the experiment on WorkflowSim platform,the results have demonstrated that ECOS can efficiently merge tasks and minimize the total cost without comprising the deadline constraint.3.In terms of priority scheduling,a heuristic task scheduling strategy based on priority and demand classification(PB-SC)is proposed to meet the different resource requirements of tasks in cloud.First,for the tasks submitted by the users,they are classified based on the resource requirement type to meet the resource requirements in the later period.Then,based on the weight of the resource,the priority of tasks is calculated and sorted.Finally,tasks are scheduled based on the ordered tasks,and greedy strategy is used to minimize the cost of tasks and meet deadlines.Compared with other comparative heuristic strategies,PB-SC strategy can effectively reduce the cost and the task completion time.