An Empirical Analysis of Internet Protocol version 6 (IPv6)

An Empirical Analysis of Internet Protocol version 6 (IPv6)

Although the current Internet Protocol known as IPv4 has served its purpose for over 20 years, its days are numbered. With IPv6 reaching a mature enough level, there is a need to evaluate the performance benefits or drawbacks that the new IPv6 protocol will have in comparison to the well established IPv4 protocol. Theoretically, the overhead between the two different protocols should be directly proportional to the difference in the packet’s header size, however according to our findings, the empirical performance difference between IPv4 and IPv6, especially when the transition mechanisms are taken into consideration, is much larger than anticipated. We first examine the performance of each protocol independently. We then examined two transition mechanisms which perform the encapsulation at various points in the network: host-to-host and router-to-router (tunneling). This is a necessary and crucial step for IPv6’s success since clear performance limitations and advantages should be well defined and agreed upon in advance before any major transitions take place. Our experiments were conducted using two dual stack (IPv4/IPv6) routers using end nodes running both Windows 2000 and Solaris 8.0 in order to compare two different IPv6 implementations side by side. Our tests were written in C++ and utilized metrics such as latency, throughput, CPU utilization, socket creation time, socket connection time, web server simulation, and a video client/server application for TCP/UDP in IPv4/IPv6 under both Windows 2000 and Solaris 8.0. Our goal was to perform an unbiased empirical performance evaluation between the two protocol stacks (IPv4 and IPv6), and how it related to the performance of the encapsulation methods utilized on identical hardware and under identical settings. Our empirical evaluation proved that IPv6 is not yet a mature enough technology and that it is still years away from having consistent and good enough implementations. The performance of IPv6 in many cases proved to be worse than IPv4, incurring an overhead much higher than its anticipated theoretical counterpart.

Ioan Raicu. “An Empirical Analysis of Internet Protocol version 6 (IPv6)”, Wayne State University, Computer Science Department, MS Thesis, May 2002, Detroit, Michigan.
Sheralli Zeadally, R. Wasseem, Ioan Raicu, "Comparison of End-System IPv6 Protocol Stacks", IEE Proceedings Communications, Special issue on Internet Protocols, Technology and Applications (VoIP), Vol. 151, No. 3, June 2004.
Sherali Zeadally, Ioan Raicu. “Evaluating IPV6 on Windows and Solaris”, IEEE Internet Computing, Volume 7, Issue 3, May June 2003, pp 51 – 57.
Ioan Raicu, Sherali Zeadally. “Impact of IPv6 on End-User Applications”, IEEE International Conference on Telecommunications 2003, ICT'2003, Volume 2, Feb 2003, pp 973 - 980, Tahiti Papeete, French Polynesia; 35% acceptance rate.
Ioan Raicu, Sherali Zeadally. “Evaluating IPv4 to IPv6 Transition Mechanisms” , IEEE International Conference on Telecommunications 2003, ICT'2003, Volume 2, Feb 2003, pp 1091 - 1098, Tahiti Papeete, French Polynesia; 35% acceptance rate.
Ioan Raicu. "A Performance Evaluation of IPv4 vs. IPv6 Forwarding using the Agere Network Processor." Purdue University, CS Dept., Networks Processors Seminar, 2002.

Webmaster Ioan Raicu: iraicu@cs.uchicago.edu

Last modified: January 06, 2008