Scalable packet classification with controlled cross-producting

Packet classification is central among traffic classification techniques that categorize packets with a traffic descriptor or with user-defined criteria. This categorization may make information accessible for quality of service or security handling on the network. To make packet classification both fast and scalable, we propose a new algorithm that combines cross-producting with linear search. The new algorithm, Controlled Cross-producting, could improve the scalability of cross-producting significantly with respect to storage, while maintaining the search latency. In addition, we introduce several refinements and procedures for incremental update. We evaluate the performance of our scheme with filter databases of varying sizes and characteristics. Specifically, we experimented with 12 different types of filter databases, whose sizes vary from 16 K to 128 K. The experimental results demonstrate the feasibility and scalability of our scheme. A comparison with the prominent existing schemes further indicates that the proposed scheme takes less time and space.