Models, algorithms, and architectures for scalable packet classification

Abstract

The growth and diversification of the Internet imposes increasing demands on the performance and functionality of network infrastructure. Routers, the devices responsible for the switching and directing of traffic in the Internet, are being called upon to not only handle increased volumes of traffic at higher speeds, but also impose tighter security policies and provide support for a richer set of network services. This dissertation addresses the searching tasks performed by Internet routers in order to forward packets and apply network services to packets belonging to defined traffic flows. As these searching tasks must be performed for each packet traversing the router, the speed and scalability of the solutions to the route lookup and packet classification problems largely determine the realizable performance of the router, and hence the Internet as a whole. Despite the energetic attention of the academic and corporate research communities, there remains a need for search engines that scale to support faster communication links, larger route tables and filter sets, and increasingly complex filters. The major contributions of this work include the design and analysis of a scalable hardware implementation of a Longest Prefix Matching (LPM) search engine for route lookup, a survey and taxonomy of packet classification techniques, a thorough analysis of packet classification filter sets, the design and analysis of a suite of performance evaluation tools for packet classification algorithms and devices, and a new packet classification algorithm that scales to support high-speed links and large filter sets classifying on additional packet fields.