A Multi-dimensional Packet Classification Algorithm Based on Hierarchical All-Match B+ Tree

Packet classification is one of the most important enabling functions for network services. The previous trie-based hierarchical search structure suffers from backtracking and address length. This paper proposes a multi-dimensional packet classification algorithm based on hierarchical all-match B+tree. By utilizing the property that the IP prefixes of a rule set are nested to each other, we develop an all-match prefix tree (APT) based on B+tree and propose a two-stage hierarchical search structure HAPT. We further propose the packet classification and rule update algorithms on HAPT. By performing multi-way search without backtracking, the proposed classification algorithm can classify packets in O(logmN) and update rules in O(mlogmN). Furthermore, the proposed algorithm is suitable for both IPv4 and IPv6. Theoretical analysis and evaluation results show that the proposed multi-dimensional packet classification algorithm improves the speed by a factor of 2-4 compared to state-of-the-art algorithm H-trie.