There have certainly been no shortage of “the sky is falling” pronouncements about the Internet. But one refrain heard often over the last 20 years is finally coming true: the Internet is running out of IPv4 addresses.
As the Asia-Pacific Network Information Center’s Geoff Huston noted in a recent posting (http://www.potaroo.net/ispcol/2011-01/addresses-2010.html), the IPv4 address space will finally be depleted in 2011, with only 33 million unallocated addresses left to serve an average annual demand in the hundreds of millions of addresses.
IPv4 depletion has long been the elephant in the room among Internet service providers. We all knew this day would come, but even over the last year there has been little momentum (and little funding) to embrace a wholesale transition from IPv4 to IPv6. That will change in 2011.
What’s the real-world impact of IPv4 address depletion? Wireless carriers and other ISPs looking to expand their services will face a choice – use 10.x.x.x private addressing space that potentially causes addressing conflicts with other 10.x networks, requires extensive network address translation, and complicates SIP-based services that rely on publicly reachable IP addresses, or rapidly migrate to IPv6. The latter solution isn’t cheap or easy, either. There is growing concern that IPv6 adoption will exacerbate scalability issues within Internet routers by rapidly increasing the size of route tables that are already straining the ability of IP routers to keep up with changes in route availability.
Aside from these two choices the only other option is creation of a market for excess IPv4 address space to enable those who have extra space to sell it to those who need it. Alternatively, governments could require those with excess space to give it back to addressing registries. Both of these create problems of their own – what if a large ISP or some other entity (Google?) buys up all available address space? What kind of herculean effort (and regulation) is necessary to recover excess space from those lucky enough to get far more than they needed in the early days of the Internet?
Fortunately, most major Internet service providers have already started implementing IPv6, though IPv6 represents a relatively small percentage of overall traffic. And providers are still struggling with not only Internet routing table issues as previously described, but in how to interface IPv4 and IPv6 networks to ensure that customers on IPv4 can reach IPv6 and vice versa.
From the enterprise perspective we are starting to see an increase in interest in IPv6, but I still can’t point to any of our clients using it in a production environment. Those who have proactively developed IPv6 plans are pursuing a dual-stack approach based on running v4 and v6 in unison, with edge gateways to enable v4 to v6 interoperability. Not only is there still not a justification for wholesale enterprise cut-over to v6, but the market still lacks support for all v4 applications, management platforms, and infrastructure. Over time, the rise of v6 services should lead to greater adoption of v6, but I’m guessing that v4 will still be around long after my kids have entered their holographic virtual office to meet with their co-workers on moonbase seven.
The bottom line: Get up to speed on IPv6. Talk to your service providers (wired/wireless) about their IPv6 plans, especially how they are addressing v4/v6 interworking and route-table scalability. Evaluate your own public facing services to ensure that they are reachable from v4 and v6, and make v6 support a requirement for all future application and network infrastructure purchases.