World IPv6 Launch - 6th Anniversary

Google is seeing close to a quarter of their access traffic over IPv6 around the world, and Facebook is reporting slightly over 20 percent. Fixed and mobile operators in the United States like Comcast and AT&T have deployed IPv6 on about 65 percent of their networks. T-Mobile in the United States is up to about 95 percent! This is happening around the world, too. We also have IPv6 implementations across a broad range of infrastructure products: routers, switches, security devices, and all major operating systems. These deployments would not be possible without that. 

Clearly, great progress has been made in the last six years.

The IETF work to develop the IPv6 protocol has been done in a series of working groups. This started with the IPv6 working group, which was renamed IPv6 Maintenance (IPv6 Maintenance) once the first protocol specifications were standardized. Operations work is done in the IPv6 Operations (IPv6 Operations). IPv6 work on more specialized link technologies happens in specific working groups like IPv6 over Networks of Resource-constrained Nodes (IPv6 over networks of resource-constrained nodes), IPv6 over the TSCH mode of IEEE 802.15.4e (IPv6 over the TSCH mode of IEEE 802.15.4e), and IPv6 over Low Power Wide-Area Networks (IPv6 over Low power WPAN).

I have been involved in IPv6 since the need for a new version of the Internet Protocol was realized in the early 1990s. This includes co-chairing the IPv6 and 6MAN working groups, and co-authoring many of the core IPv6 specifications.

About a year ago the IETF elevated the IPv6 protocol specification to the last step in its standardization process called "Internet Standard". The new specification is RFC 8200. This was a major step for the IETF and means that the core protocol specification is very stable. The Internet Architecture Board published a statement that networking standards under development take IPv6 as a given. 

The current level of IPv6 deployment means that the technical hurdles have been overcome and the implementations are mature. We would not be having this level of usage if that wasn't true. At this point, it's only a matter of will, not technological issues. There isn't a question whether IPv6 will happen, only how fast will it happen and what are the next steps. People no longer ask, "Do I need to do IPv6?" They now ask, "How do I do IPv6?"

Even with this good news, there is still a lot more to do. The next hurdle to overcome is to have sites that don't currently support IPv6 deploy it. There are still many large commerce sites, financial institutions, and small businesses who do not support it. Also, very few enterprises are using it internally. Even today, some Internet of Things and home routers don't come with IPv6 running out of the box.

However, I think these sites and devices will support IPv6 over time, and the deployment of IPv6 will continue to grow as more service providers support it and its deployment in regions that don't do it today start deploying it.

The current deployment model is called "Dual Stack", where IPv4 and IPv6 run in parallel with each other. This has worked quite well and was the right approach to starting the IPv4 to IPv6 transition. I think the next phase of transition will be to have IPv6-Only networks. In these networks, there will only be IPv6 (and no IPv4). Access to what will be the legacy IPv4 Internet will be done via translation at the edge of these networks. This is simpler and cheaper to operate as there is only one Internet protocol to configure and manage in the network. 

Service to legacy IPv4 will be similar to what is done today with Network Address Translation (NAT). Performance to IPv4 destinations will be the same, and access to sites that have IPv6 will be better. Many sites are already reporting that IPv6 service is faster. IPv6-Only is proven, for example, it is used today by mobile operators like T-Mobile. I think the next step will be to use this in enterprise. It will make enterprise networks easier and less expensive to manage. For enterprises that do this in the next few years, they may also be able to sell their legacy IPv4 addresses on the secondary IPv4 market, thus offsetting the costs of moving to an IPv6-Only environment.

I think this will be the next step in the transition to IPv6. We are not there yet, but to me it seems like an inevitable next step.