A Perspective of IP Lookup Approach Using Graphical Processing Unit (GPU)

Due to increases in communication link capacity and growth of the Internet traffic, packet processing like IP address lookup and classification becomes a major concern in the network. The packet processing performed at Switch/Router does not cope up with the growing link speed. Since Graphics processing unit (GPU) has high parallelism and more flexibility for the programmers, it can be used for solving IP address lookup problem in the Switch/Router devices. This paper proposes a variant of trie based approach to find a solution for longest prefix match (LPM) problem using GPU. In this paper, IP address database is partitioned into a different table based on first k bits of IP address, then a variant of trie approach is proposed to find the next hop. The proposed lookup approach shows 64.46% and 94.32% improvement than binary trie and BST implementations.