Study On Scheduling Algorithm Of The Independent And Associated Tasks For Cloud Computing

Task scheduling in cloud data center is the core issue of cloud computing, also thekey technology for cloud computing’s large-scale application and system performanceimprove. Advanced task scheduling have a great significance for cloud serviceproviders to improve the efficiency of computing resources, save energy, improveresource sharing, and reduce operating costs deserves further systematic study.The main work of this paper:(1) Studied the key technologies of cloud computing, analyses several current hotresearch task scheduling algorithm in academia detailed and comparatively; andresearched the independent and associated task scheduling algorithm in cloudcomputing deeply. we found some deficiencies, and proposed the correspondingmethod.(2) For the independent task scheduling, we consider user’s satisfaction and cloudservice provider’s avails synthetically, and then propose a merge scheduling algorithmwhich combine genetic algorithm with particle swarm optimization. Firstly, we classifythe virtual machines in cloud computing system, and introduce the degree oftask-resource satisfaction, resource comprehensive performance. Secondly, we optimizeparticle initialization operations to improve the quality of the initial particle in particleswarm optimization. Thirdly, in order to overcome the defect of particle swarmoptimization’s easy falling into local optimal solution, we integrate genetic algorithm toextend the search space. Finally, compared with other scheduling algorithm,experimental on cloud simulation software Cloudsim show that this paper algorithm isan effective scheduling algorithm which can effectively improve user’s satisfaction andincrease cloud service provider’s gain.(3) For the associated task scheduling, under the fundamental of studying classicalscheduling algorithm we propose a new one based on task priority and task replication.In heterogeneous environments of cloud computing, critical path in DAG taskscheduling has lost its meaning to the most urgent task. In this paper, when determiningthe priority task, we consider factors such as the task communication cost, the taskcomputing cost, the task out degree and the branching values of task comprehensively.In the processor selection stage, we use the means of replicate redundant tasks to advance the time of task’s beginning execution. Through these two aspects improving,both theoretical and experimental analysis show that this proposed algorithmcomparison of existing scheduling algorithms, reducing the cost of communicationdelay effectively, shorten the complete time of the DAG task.In this article, we study the independent tasks and the associated tasks schedulingalgorithm under the cloud computing environment. The proposed scheduling algorithmof the independent task is more suitable for the cloud data center’s large throughputenvironments; the proposed scheduling algorithm of the associated task is more suitablefor communication-intensive task scheduling. Research on the subject provides areference and new ideas to solve cloud computing task scheduling problem.