After my previous article about Cisco LISP and the IPv6 transition, I fully expected to hear about all the ways that LISP could improve things. I had argued the merits of tunnel services over LISP as a transition tool. Instead, I was pleasantly shocked when I got an email from Fred Baker, a Cisco Fellow who co-chaired the IPv6 Operations Working Group in the IETF. Fred gave the first Cisco Live presentation I ever attended back in 2006, which was based in large part on his work with RFC 4192, Procedures for Renumbering an IPv6 Network without a Flag Day. It was my first stop down the IPv6 highway.
Fred wanted to talk to me a bit about my stance on IPv6 -- how it's not being widely deployed outside LANs that have it enabled by default. He gave me some interesting numbers about IPv6 adoption, which indicate that we may not need LISP or tunnel services to aid the transition from IPv4. Based on graphs from the Réseaux IP Européens Network Coordination Centre (RIPE), around 17% of the Internet is advertising IPv6 prefixes to the global routing table. There are more prefixes being originated from the Asia-Pacific Network Information Centre (APNIC), but that makes sense when you realize they are in the final stage of depletion for the remaining IPv4 prefixes.
The more interesting number comes again from RIPE when you cross reference the number of IPv4 prefixes that are also announcing IPv6 prefixes. Almost 70% of the IPv4 address space is being announced by networks that also announce IPv6. That's pretty impressive. When you break it down by the size of the IPv4 network, it gets even better: The majority of networks with more than 100,000 IPv4 addresses are announcing IPv6 as well.
Fred also pointed out that a growing number of non-transit BGP networks are announcing IPv6 prefixes. That means it isn't just providers; large enterprises have already begun to transition to IPv6 as well. Whether it be for announcing their presence to the greater IPv6-enabled Internet or just to do something with all those Windows and Mac systems already running IPv6 is anyone's guess. The key is that they've started. They aren't waiting for the ISPs to worry about providing access or hoping that solutions like Carrier Grade NAT (CGN) are going to keep the IPv4 address pool from depleting all the way.
Enterprises are already adopting IPv6 for the future. Consumers on the other end of the wire just want things to work, whether for Web browsing or other services. However, as more consumers start using technology that allows for remote access, it will be critical to provide a direct connect capability. Fred told me a story about installing a home security camera system. People really like installing an app on their phone or tablet and getting on-demand access to view their home to ensure it is safe. How will that work through NAT, CGN, or whatever else comes on top of all of that?
Enterprises are already seeing problems with deploying services across NAT. It makes sense when you realize that more and more of them are announcing IPv6. Software vendors also are using IPv6 more and more. Take a look at what happens in Exchange 2013 when you remove the IPv6 binding from the network adapters. The smart folks have already woken up and realized that IPv6 is here to stay -- they aren't waiting for LISP to fix their problems. Instead, large companies are pulling themselves up by the bootstraps and building the IPv6 Internet. They just aren't filing press releases when they turn up a new prefix.
Are you announcing IPv6 prefixes to the global routing table? If so, are you doing it as a test? Or are you enabling services for the growing number of consumers and commercial customers that are connecting with IPv6? Share your comments below. I would love to hear how many of my readers are posting comments from an IPv6 address.