A Fast and Scalable IP Lookup Scheme for High-Speed Networks

  • Authors:
  • Wen-Shyen E. Chen;Chung-Ting Justine Tsai

  • Affiliations:
  • -;-

  • Venue:
  • ICON '99 Proceedings of the 7th IEEE International Conference on Networks
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

The past few years have witnessed the advent of the World Wide Web and the emergence of multimedia applications over the Internet. With the amount of the traffic doubling every three months and the introduction of the LAN equipment of ever increasing speeds from 100Mbps to Gbps, high-speed routers that process millions or 10's of millions of packets per second are needed to meet the future demands. One of the key design issues for the next generation IP routers is the IP lookup mechanism. For each incoming IP packet, the IP routing requires to perform a longest prefix matching on the address lookup in order to determine the packet's next hop. Currently, the process is done in software and has become a major bottleneck of the router performance.In this paper, we propose a high-speed IP lookup scheme for the Best Matching Prefix (BMP) by using {\em forwarding tables} consisting of Prefix Information Tables (PITs) and Lookup Tables (LTs) that provide guidelines for efficient search. The scheme scales very well as the sizes of the address and the routing table increase. For IP lookup in IPv4, the scheme needs 1 memory access in the best case, and 2 memory accesses plus one hash in the worst case to locate the BMP, if one exists. It requires only 540 KBytes to 630 KBytes of memory space when about 42,000 routing tables entries in the backbone are logged for simulation. When 50ns DRAM is used for the forwarding tables, the scheme offers lookup speed of 10 millions packets per second (10 MPPS). The lookup speed can be improved linearly with the speedup of the type of memory used. Further, it does not need complex compression mechanisms to reduce the memory requirement. Its operation is simple, lending itself for easy hardware implementation and therefore is a good candidate to be used in the next generation IP routers.