| As the focus studies of domestic and foreign research institutions, cloud computing is an emerging computing model. It follows distributed computing, parallel computing and grid computing. It is also a resource providing model based on service, making various kinds of resources virtual into various services. Task scheduling technique has been one of its hot topics. The user groups for cloud computing are the general public. Due to the extremely extensive application situations, huge tasks and large amount of data are almost always generated. Thus, it is a vital issue how to effectively classify cloud resources so as to achieve highly efficient schedule for tasks.The target of task scheduling technology in cloud environment is to abtain a reasonable match for tasks and resources, however, traditional scheduling algorithms mainly consider overall service resources and only aim to shorten task complete time while choosing resources for tasks, which can not really satisfy QOS requirements for users, a Task Scheduling algorithm based on Non-Uniform Granularity Classification(TSNUGC) was proposed in this paper. Our proposed algorithm uses the idea of non-uniform granularity to do classification preprocessing for cloud resources, by calculating and analyzing task preference, scheduler selects the resource whose performance is much better and the load that keeps balanced in corresponding resource category. Scheduling strategies search for, on one hand, minimizing completion time for tasks and enhancing resource matching rate, on the other hand, balancing resource load and improving system throughput. Simulation results show that the proposed algorithm can well meet user requirements. It not only minimizes the complete time while holding resource load in a relatively balanced state, but also enhances resource matching rate and system throughput, leading to the significantly improved service performance for cloud system. |
PDF Full Text Request