IPv6 Decision Time

I build Internet infrastructure and data center networks in my day job. Then I go home and use IPv6.

This has been my status for many years (ever since I started working on IPv6 way back in 2005). By extension this has also been the status of IPv6. IPv6 has had a long, painful journey toward mainstream adoption, and I still think it has a long way to go. But I can tell you this: if you are not paying attention to IPv6, you are going to hurt your company and your career.

Granted, I am an IPv6 zealot, evangelist, or cheerleader (whatever label you want to use) but I think I have a very solid argument. You can disagree about timelines but you can’t ignore IPv6 anymore. Why?

Because we’ve come to the end of IPv4 address space. IANA ran out of IPv4 in 2011, APNIC and RIPE effectively ran out in 2011 and 2012 respectively, and ARIN is due to run out in 2014. This isn’t an end-of-the-world event (IPocalypse) because the service providers that get the IPv4 addresses from these organizations still have an inventory of IPv4 to use. The real question is, what happens when they do run out of IPv4?

I see two possible paths forward. Neither are painless, but one is more painful and costly than the other.

The first path is for service providers to deploy IPv4 conservation methods. They reclaim unused IPv4 address space and implement Carrier Grade NAT (CGN) solutions to hold onto as much public IPv4 address space for as long as possible. At best, this might buy us a few more years, but the pain will be extremely high.

Even if carriers implement an extremely aggressive IPv4 addressing recovery plan (mandating 99% utilization of public IPv4 allocation) there still aren’t enough IPv4 addresses to accommodate the growing demand for them. Geoff Houston tracks available IPv4 address space, and has fascinating stats here.

IPv4 conservation methods will be expensive for the carriers, and the cost will get passed down to customers, including enterprises, medium and small businesses, and consumers. IPv4 and CGN become very expensive because the carriers will have to deploy hardware that performs NAT on behalf of their customers.

Customers end up with shared IPv4 addresses. This means things like peer-to-peer applications, Web sites that use a lot of sessions (Google Maps comes to mind) and VPNs become much more complicated to run, if they work at all. That is a not a solution that works for the growth of the Internet and doesn’t meet the needs of enterprise customers.

And IPv4 conservation and CGN has limits, because eventually you simply run out of IPv4 addresses. You can only NAT so many times before it becomes impossible to support. Then what?

The second path is a transition. It requires an investment in training people, testing applications, and moving a lot of resources from a legacy networking protocol. But you can do this over time--it doesn’t have to happen overnight.

Will it be painful and cost money? Yes. But organizations that choose the first path will have to move to IPv6 at some point anyway. It doesn’t make sense to pay twice.

So how do enterprises avoid unnecessary pain? You invest now in IPv6 and choose service providers that give you a native IPv6 option. That way your company plan can decide the timeline to deploy IPv6. Get in the driver’s seat and be in control of your Internet experience.

What’s your IPv6 strategy? I can help you answer that question at my Interop session “Getting Serious About IPv6: Go Big Or Go Home.”