Path Provisioning for Fibbing Controlled Load Balanced IP Networks

Abstract

With the help of distributed link state routing protocols, IP networks are highly reliable and robust. However, routing in IP networks are relatively inflexible. Flows can follow only shortest paths toward their destinations. In contrast, Software Defined Networking (SDN) offers fine-grained control over routing, at the expense of controller overhead, failover latency, and deployment challenges. The best choice is to combine both approaches to achieve “central control over the distributed route computation” paradigm. Recently a new idea called Fibbing is proposed to realize flexible fine-grained control in conventional IP networks. Fibbing as an alternative loop-free destination based solution brings the flexibility for traditional networks without changing router configurations and/or protocols. The Fibbing controller can steer a traffic flow to follow any specific route. However, how to achieve traffic engineering through dynamic traffic monitoring and route computation is still an open question. In this paper, we propose a framework to achieve an optimized load balancing solution for a Fibbing controlled network. In our system, there are a light weight traffic monitoring module, a demand estimation module, and an optimization engine for routing path computation. The system periodically determines a set of routing paths to the Fibbing controller for path provisioning. In order to evaluate the performance, we have implemented the whole system in a testbed. The experimental results show that the average time for file transfer is significantly reduced with the support of the proposed system.