Joint Optimization of Flow Latency in Routing and Scheduling for Software Defined Networks

Meng Shen,Liehuang Zhu,Mingwei Wei,Qiongyu Zhang,Mingzhong Wang,Fan Li
DOI: https://doi.org/10.1109/icccn.2016.7568535
2016-01-01
Abstract:Software Defined Networks (SDNs) decouple control plane from data plane and enable fine-grained traffic management by a logically centralized controller. Reducing the flow latency is of great importance in traffic management, which benefits both service providers and end users. Routing design and flow scheduling are typical ways to improve the flow transmission efficiency. However, existing studies usually consider them separately, due to the complexity of joint consideration.In this paper, we combine the routing and scheduling together and propose a latency-aware routing scheme with bandwidth assignment, which can efficiently reduce the flow latency with a moderate complexity. In the routing design, we utilize the global flow information to reduce both the latency of the newly arrived flow and its interference with existing flows in the network. Given flow forwarding paths determined by routing, the flow scheduling dynamically reallocates the bandwidth to all flows so as to further reduce the total flow latency. Experimental results show that our scheme outperforms the scheme currently available in OpenFlow, with an improvement of up to 60% on flow efficiency and a higher percentage of flows that meet their deadlines.
What problem does this paper attempt to address?