A method of processing inbound IP packets with a routing table

As the Internet continues to spread and equipment performance improves, the typical operation has been to assign multiple Internet Protocol (IP) addresses to one node. At this type of node, whether the node is the destination of the received packet must be efficiently determined. Conventionally, a Berkeley Software Distribution (BSD) system used an algorithm that performs a linear search for the address of the node. As described in this paper, the Widely Integrated Distributed Environment (WIDE) Project conducted R&D on an improved method using a routing table algorithm. Later an original improvement using a hash table was introduced into Internet Protocol Version 4 (IPv4) implementation in NetBSD and BSD/OS. In this paper, we explain these three methods and evaluate their performances using actual equipment. From the evaluation results, a decision method based on the routing table or the hash table is clearly an improvement over the conventional technique based on a linear search. In addition, the performance of the decision method using the routing table algorithm does not degrade even if the number of addresses increases similar to a conventional routing lookup.