Longest Prefix Matching with Pruning

Lingtong Liu,Jun Hu,Yibo Yan,Siang Gao,Tong Yang,Xiaoming Li

DOI: https://doi.org/10.1109/hpsr.2019.8808125

2019-01-01

Abstract:Nowadays, an explosive increase of the size of Forwarding Information Base (FIB) in backbone router makes IP lookup a challenging issue. An effective solution is to use Bloom filter. Because of its false positive, to boost longest prefix matching (LPM), sophisticated Bloom filter-based algorithms were proposed, which makes FIB updating extremely difficult. To address this issue, in this paper, we propose a novel IP lookup algorithm which prunes unneeded prefix probes incurred by false positive and supports FIB updating. Experimental results show that our algorithm can save approximately 7 ~ 24% on-chip memory space than the well-known PBF (longest Prefix matching using Bloom Filters) while keeping the same average number of probes of 1.003 per IP lookup. In other words, given a limited on-chip memory budget, our proposed algorithm can handle FIB 7 ~ 24% bigger than PBF. The source code can be found in [1].