Posted by: Jeff Young
As the Internet nears doomsday (the day when the last available Internet address is assigned) those of us in the networking community might want to brush up on a few facts and even prepare an elevator speech for the few times when we're caught alone with a well-meaning but 'informed-by-NBC' type coworker. Of course that coworker might turn out to be the boss. As we near doomsday rest assured that the popular media will start picking up the story and as always, certain outlets will get it wrong.
Sure, we're about to run out of IP addresses, the best prediction we have for doomsday is a running prediction by Geoff Huston of the Asia Pacific Network Information Center (APNIC). Today's prediction is the 28th of January, 2012 (see http://www.potaroo.net/tools/ipv4/index.html). Huston's site is a wealth on information for those of use who need to brush up on the issue, he and Tony Hain of Cisco are the two researchers most well known for coverage of this topic.
But the Internet isn't really in imminent danger, or is it? Let's just say that the future of the Internet as a reliable, connected, global communications medium is in doubt. The Internet isn't going to stop working in 2012, but those who connect to the Internet after that date may find that it's not their father's Internet anymore.
The issue is growth. How can the Internet grow if the pool of IP addresses is gone? Well the good people in the IETF started working on this problem back in 1993 and the answers they came up with led us to the situation in which we now find ourselves. The first answer to the problem was to adapt IP address assignment and routing to better fit the size of enterprises who needed IP address space. Classless InterDomain Routing (CIDR) stretched the IP address space we had for these 15 years. The second answer was a bit controversial, but an answer just the same. Network Address Translation (NAT) allowed enterprises and service providers to multiplex IP addresses or use different IP addressing inside an enterprise than was used on the Internet. NAT hasn't been very popular with many in the IETF, but ironically NAT is now at the center of an effort within the IETF to make IPv4 backwardly compatible with IPv6.
The third answer to the problem, is of course, IP version 6 (IPv6). IPv6 is the new Internet protocol. It was intentionally designed to be incompatible with IPv4 and although we the Internet should have 'transitioned' to IPv6 by now, it hasn't. Hence the concerns surrounding doomsday.
We should all be running IPv4 and IPv6 computers at this point – well, yes we are because all modern operating systems support both protocols. We should have connections to an IPv6 as well as an IPv4 Internet. Many of us have those connections as well, although we may not know it – some of these operating systems bring up tunnels on our behalf unless we disable that 'feature.' All of the content that we find on the Internet should be available on IPv4 as well as IPv6 servers – wrong. In fact, we've probably spent so much time fighting the notion that we need to 'transition' to IPv6, which costs real money, that we've neglected the obvious.
In a few short years there will be IPv6-only hosts out there on the Internet and they'll want to connect to the services we, as enterprises, provide!
So if you need the elevator speech, you finally have something intelligent to say about IPv6 when the boss asks, here it is:
“Ignore the smoke, there is no fire just yet. However, all of those late comers to the Internet are going to want access to our external services (web, email, VPN, and so on) that we, as an enterprise have provided on the Internet for years. That means we need to budget and plan to extend these services over IPv6 in two years time.”
Sure there is a move afoot to solve the IPv6 doomsday problem in the IETF. It's largely based on NAT (or more precisely NAT-PT) and it has already fallen flat on its face once. Sure the Regional Internet Registries (those guys who dole out IP addresses) might relax their prohibition on IP address transfers and allow local markets for IPv4 addresses to form. Read Geoff Huston's site, we're consuming 12 /8 networks per year and we have 38 of those networks left, there simply isn't enough space to continue the Internet's growth on IPv4.
To sum it up, the time for debate is gone. IPv6 is a reality and you should start planning to adopt it in your external facing data centers. No one is telling you to move you're internal enterprise backbone to IPv6 (although you might consider the protocol for small pilot projects in inventory management, surveillance and so on). In fact, you'll probably be running IPv4 on your internal networks the same way we ran IPX, XNS, AppleTalk, and other protocols long after the Internet Protocol had become the defacto standard.
Comments