The Research Of Task Scheduling Based On Cloud Environment

Cloud computing uses virtualization technology to virtual data center’s hardware resource into a dynamic and scalable virtual resource pools, providing high availability, high scalability, high security services for client via the Internet. Due to a large number of different types of tasks and a lot of virtual resources, a good task and resource matching strategy can effectively improve the operating efficiency and load balancing of the cloud computing system.Aiming at the single users’ Qo S requirement, low system resource utilization and other problems of task scheduling in cloud computing environments, the paper carries on the investigative work, The main work are the follows aspects:(1) As cloud computing environment is virtualized, heterogeneous environments and scalability characteristics,resulting in a quite different between cloud workflow and traditional workflow in architecture, the paper presents a workflow architecture based on cloud computing environments, meanwhile, the architecture of various levels have been described in detail.(2) The paper proposes a cloud workflow scheduling algorithm based on the deadline constraints(Deadline Partial Critical Path, Deadline-PCP) to meeting the users task deadline constraints and make the minimum cost of task execution. First, it dynamic divides the user defined global deadline for each task, making each task has a child deadline. Then schedule the task to the virtual machine which has minimum cost and satisfy the task’s child deadline.It also puts forward the concept of critical path, assigning the tasks of critical path to the same virtual resources, this can effectively reduce the transmission time to shorten the whole task completion time. Finally, the Cloud Sim simulation tool is extended by simulating a DAG Builder, which randomly generates input instances, overriding key classes and some other methods. After that the platform can support the workflow scheduling simulation under the cloud. The experimental results show that the algorithm of this paper has greatly improved in terms of task execution time and cost.(3) Aiming at the single users’ Qo S requirement, low system resource utilization and other problems of task scheduling in cloud computing environments, the paper proposes a multi-dimensional Qo S cloud workflow scheduling algorithm based on critical parent task replication(Multi-dimensional Qo S on Critical Parent Task Replication, Qo S-CPTR). Firstly,build a cloud computing model, a resource Qo S model is put forward. Secondly, the criterion that basing on the user preference to evaluate the performance evaluation of virtual resources is proposed, the task is performed, virtual resources with higher comprehensive or more in line with the user preferences are chosen to allocate.Simultaneously, in order to shorten the execution time, this algorithm duplicated a critical parent task on idle time of this virtual resource. The experimental results show that this algorithm can improve system resource utilization, improve user satisfaction and reduce the user cost.