Comparison Between IPv4 to IPv6 Transition Techniques

TL;DR

This paper experimentally compares IPv4 to IPv6 transition techniques, focusing on their performance, user experience impact, and resource usage in scenarios with limited IPv4 addresses and IPv6-only networks.

Contribution

It provides an empirical evaluation of DS-Lite, 464XLAT, MAP-E, and MAP-T transition techniques under realistic usage conditions and network performance metrics.

Findings

DS-Lite and 464XLAT show lower latency and jitter.

MAP-E and MAP-T have higher processor usage.

Transition techniques impact bandwidth and user experience.

Abstract

The IPv4 addresses exhaustion demands a protocol transition from IPv4 to IPv6. The original transition technique, the dual stack, is not widely deployed yet and it demanded the creation of new transition techniques to extend the transition period. This work makes an experimental comparison of techniques that use dual stack with a limited IPv4 address. This limited address might be a RFC 1918 address with a NAT at the Internet Service Provider (ISP) gateway, also known as Carrier Grade NAT (CGN), or an Address Plus Port (A+P) shared IPv4 address. The chosen techniques also consider an IPv6 only ISP network. The transport of the IPv4 packets through the IPv6 only networks may use IPv4 packets encapsulated on IPv6 packets or a double translation, by making one IPv4 to IPv6 translation to enter the IPv6 only network and one IPv6 to IPv4 translation to return to the IPv4 network. The chosen techniques were DS-Lite, 464XLAT, MAP-E and MAP-T. The first part of the test is to check some of the most common usages of the Internet by a home user and the impacts of the transition techniques on the user experience. The second part is a measured comparison considering bandwidth, jitter and latency introduced by the techniques and processor usage on the network equipment.