Research On Task Scheduling Of Cloud Computing Based On Service Quality

As a new business model, cloud computing provides services to the users through the virtualization of a series of dynamic scalable resources. According to the quality of service (QoS), cloud platform distributes appropriate resources to tasks, which can maximize the customers’ satisfaction and resources’utilization. In the face of a large number of users and tasks, the performance of cloud computing platform plays a decisive role, and the level of task scheduling strategy directly influences the performance of cloud computing platform. Therefore, how to schedule tasks according to users’quality of service is one of the most important problems.Quality of service is the defined features and characteristics of the services, it is the minimum level of service to achieve customer satisfaction. This paper provides a new method for task scheduling in cloud computing. Numerous researchers studied and established model and strategy of task scheduling to meet the quality of service constraints. Nonetheless, there exist some problems, for instance, lacking reliable measurement method of priority and ignoring the dynamic scheduling mechanism under the influence of uncertain events. Based on satisfaction of quality of service and complete time of tasks, this dissertation researched mainly evaluation method of task priority, model of task scheduling and dynamic scheduling mechanism. In summary, the paper mainly completes the following work:(1) By analyzing the influence factors of task priority in cloud computing, the paper establishes a task level evaluation system based on the deadline, leisure time, value of tasks and the type of user. This paper calculates the priority by fuzzy comprehensive evaluation method.(2) Combined with the existing types of resources, the paper clusters the tasks based on five features:computation ability, communication ability, storage ability, safety and reliability. According to the clustering results, this paper assigns tasks to the same kind of resources, which reduces the search scope.(3) Quality of service is measured by aspects (i.e., performance and economy). The membership function is used to quantify the differences between the actual completion level and expected completion level of tasks. With considering task priority, this paper establishes a task scheduling model, and designs a dual fitness genetic simulated annealing algorithm to solve the model.(4) We put forward a dynamic task scheduling strategy with feedback, which monitors and feeds back the properties and status of tasks. This paper forms a loop feedback scheduling process. Finally, combined with interference management, this paper adjusts the scheduling scheme dynamically, and reduces the effects of uncertain events.