CuckooFlow: Achieving Fast Packet Classification for Virtual OpenFlow Switching by Exploiting Network Traffic Locality

Abstract

The growing trend of network virtualization leads to a widespread deployment of virtual switches in virtualized environments. However, virtual switching is confronted with great performance challenges regarding packet classification especially in OpenFlow-based software defined networks. This paper takes an insight into network traffic locality from the perspective of exact flows and wildcard flows, and propose a fast packet classification scheme CuckooFlow. In particular, we devise an active exact flow cache with cuckoo hashing to predict the respective flow entry for each arrived packet. Subsequently, we formulate its cache hit rate based on packet-in-batch arrivals, and average search length of the packet classification scheme. Finally, our proposed CuckooFlow scheme is evaluated with real network traffic traces by experiments. The experimental results indicate that our proposed scheme outperforms traditional ones Open vSwitch and CuckooDistributor in terms of cache hit rates and average search length, and contributes to promote virtual OpenFlow switching performance.