Proceedings of the IFIP Int'l Conference on Networking (Networking), pp. 756~769, Aachen, Germany, May 2009.
Building a high performance IP lookup engine remains a challenge due to increasingly stringent throughput requirements and the growing size of IP tables. An emerging approach for IP lookup is the use of set associative memory architecture, which is basically a hardware implementation of an open addressing hash table with the property that each row of the hash table can be searched in one memory cycle. While open addressing hash tables, in general, provide good average-case search performance, their memory utilization and worst-case performance can degrade quickly due to bucket overflows. This paper presents a new simple hash probing scheme called CHAP (Content-based HAsh Probing) that tackles the hash overflow problem. In CHAP, the probing is based on the content of the hash table, thus avoiding the classical side effects of probing. We show through experimenting with real IP tables how CHAP can effectively deal with the overflow.