期刊名称:International Journal of Computer Science and Network Security
印刷版ISSN:1738-7906
出版年度:2016
卷号:16
期号:5
页码:67-76
出版社:International Journal of Computer Science and Network Security
摘要:Software Defined Network (SDN) has shown substantial benefits over the legacy network and fueled the implementation of a variety of innovative and intelligent applications on SDN Controller. However, these applications put the performance of SDN controller under question since most of these applications excessively demand computing resources of SDN controller resulting in increased flow processing delay, and consequently performance of the controller reduces. Accordingly, in this paper, we investigate the potential of Graphics Processing Unit (GPU) to address this performance issue by accelerating the computationally/memory intensive tasks of SDN applications on GPU. More specifically, in this paper, we are considering SDN based traffic load balancing application in a large scale Data Center Network (DCN) as a case study to see how GPU based approach can improve performance of SDN controller. We offload computations of traffic load balancing application on GPU and analyze the performance gains in terms of throughput, latency and speedup. The preliminary performance evaluation results show that GPU has an impressive capability to improve performance of SDN controller.
关键词:SDN; OpenFlow; Controller; GPU; GPGPU Computing; CUDA