Empirical Performance of IPv6 vs. IPv4 under a Dual-Stack Environment

As we continue to see the increasing global-scale deployment of Internet Protocol version 6 (IPv6) networks, it is equally important to evaluate the performance of these IPv6 networks. In this paper, we present comprehensive empirical measurements of the IPv6 network performance from an end- user´s perspective. In particular, by sending probing traffic from our dual-stack IPv6/IPv4 testbed to over 2,000 dual-stack hosts worldwide, we quantify the performance differences of using IPv6 vs. IPv4, in terms of various network metrics like network connectivity, hop count, RTT, throughput, operating systems dependencies as well as the address configuration latency. We also investigate the performance impact of using IPv6 tunneling brokers instead of native IPv6 services. Whenever possible, we also compare our measurement results with previously published ones to reflect on the progress of IPv6 deployment/performance improvements in the past few years. In general, our results indicate that in these past few years, the IPv6 backbone has improved considerably to provide more than 95% network connectivity to various IPv6 sites. The end-user-perceived performance offered by some IPv6 tunnel brokers are also comparable to that of native IPv6 services.