Design of a Load-Balancing Middlebox Based on SDN for Data Centers
Speaker(s) : Jin Zhao (Fudan University, China)
Modern cloud-based applications are becoming more communication-intensive, posing a challenge for data centers to supply sufficient bandwidth. Current data centers are not optimized for such cloud-based software services. In practice, due to the variability of traffic load, data centers may suffer from lower throughput, as well as higher latency, dramatically reducing quality of service (QoS). This paper addresses this challenging problem by introducing a programmable middlebox that can distribute traffic more evenly. The middlebox is based on a Clos network design and uses Software Defined Networking (SDN) to enhance bandwidth utilization while guaranteeing QoS. The SDN controller of the middlebox collects information from switches and servers to learn the traffic distribution and server loads and to perform load balancing accordingly. When implemented in a data center, the middlebox can greatly improve bandwidth utilization and reduce latency up to 50%. Since the middlebox does not rely on any specific feature of the data center, it can readily be deployed in current data centers.
Bio: Jin Zhao is an associate professor at Fudan University. He received the B.Eng. degree in computer communications from Nanjing University of Posts and Telecommunications, China, in 2001, and the Ph.D. degree in computer science from Nanjing University, China, in 2006. He joined Fudan University in 2006. He stayed at University of Massachusetts Amherst for 1 year as a visiting scholar in 2014. His research interests include software defined networking, media streaming and network coding theory. He is a member of IEEE and ACM.