Flowcache: A cache-based approach for improving SDN scalability

Today, the transition of traditional networking model to SDN-type architectures poses several major challenges. In typical SDN settings, the routers and switches frequently generate requests to the controller which ensures the proper and efficient operation of the network. Increased workload on the controller results in larger control plane response times, which, in turn, leads to a delayed response to the data plane events. As a result, the controller can become a major performance bottleneck which will negatively affect the performance of the overall system. This problem is important in access and edge networks where a controller is expected to serve a large number of remote switches. In this paper, we propose to address this problem by applying the principle of caching to the control plane of the SDN framework. In particular, we propose to augment an SDN architecture with a transparent layer in between the controller and switch, referred to as a flowcache. Flowcache acts as a software cache for the control traffic, by temporarily storing the content of recent flow table entries modified by the switch. This results in a significant reduction of the access time for future requests of similar flows. We analyze different design choices for the flowcache, analyze its properties, and evaluate the benefits of introducing a flowcache in an SDN architecture.

I. Introduction

Software Defined Networking (SDN) has recently gained popularity within the networking industry. The SDN model provides flexibility in both designing and managing communication networks with a logically centralized controller. In SDN architectures, the controller maintains a centralized view of the underlying network and installs flows in the forwarding elements (switches) to route traffic along specific paths. This approach allows network administrators to configure networking devices to meet the requirements of the network applications.