Everyone is wondering when IPv6 will actually be deployed in earnest on the global Internet. While there are a handful of enterprises that have extensive internal IPv6 deployments, the ratio of IPv6 to IPv4 traffic in the global Internet is still very small (See “World IPv6 Day: Final Look and “Wagon’s Ho!”). I have a theory about what will trigger significant deployment of IPv6 on the global Internet, which I’m presenting in this post. Events can certainly occur that will skew IPv6 deployment and contradict this theory, but no one I’ve broached this theory with has outright dismissed it.
It is obvious that the one thing that can tip IPv6 into widespread deployment is the situation where IPv4 addresses simply become unavailable.
It is unlikely that a business that relies upon an Internet resource, such as a retail web site, will consider allowing their web site to operate without an IPv4 address. At this stage of IPv6 deployment, the maintainer of a data center or a hosting provider will never tell its paying customers that they can only use IPv6 addresses. This would be akin to a hosting provider telling its customer businesses that their web servers will only be reachable by users with computers running Linux operating systems. The data center or hosting provider will do whatever it can to make IPv4 addresses available to any customer. This may require some interesting networking and implementations of network address translation (NAT), but for the data center or hosting provider to not offer IPv4 connectivity would be financial suicide. If a hosting provider can’t provide content via IPv4, customers will look for another hosting provider.
So, I’ve established that it is unlikely that a provider of web content will only offer their content via IPv6. It just doesn’t make sense that web content providers will be the catalyst in widespread IPv6 deployment.
It seems more likely that a large service provider (perhaps a home or cellular Internet provider) will reach the state where they have new end-user customers, but no more IPv4 addresses to hand out. The only practical option will be to provide new end-user customers with good network connectivity via IPv6. It is even initially possible that the service provider could create a strategic advantage for its business customers via indirectly providing a “walled” web environment consisting of sites that can be accessed via IPv6.
When the service providers’ users are given only IPv6 addresses, businesses, such as retail web sites, that wish to reach these customers will have to find some way of reaching them via IPv6. The business must natively enable IPv6 on their content/web servers, using either dual-stack capable load sharing devices or translators. After all, businesses will not want to turn their back on potential profits from IPv6-only customers.
John Spence, vice president of IP Services at Nephos6, thinks it is likely a provider that only hands out IPv6 addresses to users will want to provide DNS64/NAT64 capabilities. This would give IPv6-only users access to “low complexity services such as IPv4-only websites that will remain IPv4-only for an extended period – not the Googles or Facebooks of the world (which will go dual-stack soon), but the local PTA or Chamber of Commerce website,” Spence said. (DNS64/NAT64 refers to a promising transition mechanism where DNS64, “a mechanism for synthesizing AAAA records from A records” ¹, is used with NAT64, “which allows IPv6-only clients to contact IPv4 servers using unicast UDP, TCP, or ICMP”².)
In a utopian world, network content providers will quickly learn that maintaining translation devices has its share of problems and that moving to a fully dual-stack IPv4/IPv6 environment will be much less resource intensive. Content providers that have tested and implemented a fully dual-stack IPv4/IPv6 environment will have a distinct advantage over providers who discover after the fact that they are losing money because they can’t reach IPv6-only customers.
Certainly, I may be looking at the networking world with IPv6-colored sunglasses. I am a believer that with IPv4 address exhaustion finally upon us, that the global networking community will ultimately decide to deploy IPv6. I think that IPv6 growth will only be driven by a condition where businesses find that they must IPv6-enable their Internet services to reach IPv6-only customers.
As already stated, this is my theory about how IPv6 will become broadly deployed within the global Internet. I welcome suggestions about flaws in this theory or alternate theories that seem more feasible.
1. IETF RFC 6147, “DNS64: DNS Extensions for Network Address Translation from IPv6 Clients to IPv4 Servers,” Bagnulo, M., Sullivan, A., Matthews, P., and van Beijnum, I., April 2011
2. IETF RFC 6146, “Stateful NAT64: Network Address and Protocol Translation from IPv6 Clients to IPv4 Servers,” Bagnulo, M., Matthews, P., and van Beijnum, I., April 2011