DHCPv6 Is Getting to Be a Mature Technology

By: Alain Durand

Date: July 27, 2008

line break image

About three years ago, when Comcast started to look at production-grade DHCPv6 servers, not much was available. Sure, there were a few proof-of-concept codes here and there, but nothing we would dare use in production. Things have changed quite a bit since then.

Back in early 2007, only a few products, both commercial and open source, were available, each at various stages of maturity. After an initial attempt at interoperating them, it quickly became apparent that there were hiccups. Clearly, more testing and polishing were necessary. In the tradition of early connectathon and interop events, we decided to organize a DHCPv6 bake-off that would bring together implementers of all of the available codes. They would be locked in a room for a couple days and asked to build real networks in which all products would interoperate seamlessly.

The first bake-off was held in March 2007 at the RIPE NCC (the event was reported in the IETF Journal, Volume 3, Issue 1). Participants were eager to repeat the event by the end of the year, so another bake-off was performed in Vancouver, Canada, just after IETF 70. While the implementations were improving, the participants asked for a third bake-off, which was held in Philadelphia right before IETF 71.

Today, more than 20 separate implementations have been proved able to interoperate. The technology is maturing and is beginning to be deployed. Furthermore, both commercial and open-source servers are readily available. For example, the IETF 71 IPv6-only network was for the first time offering stateful DHCPv6 address allocations. Not only did the staff report no problems with it; they also characterized it as a simple upgrade from the server code that had previously been used for DHCPv4.

Many operating systems are able to take advantage of the technology and are now integrating (natively) a DHCPv6 client. The next step is to get DHCPv6 clients and servers integrated into home routers, which is essential for distributing IPv6 addresses, prefixes, and other configuration data to residential broadband customers. So far, this remains the missing link, since very few of those devices currently support IPv6.

To all of the participants of the three bake-offs, we thank you for your decisive contribution to making this technology a reality and for serving as cornerstones of the deployment of IPv6.