Hey there! Thanks for dropping by Theme Preview! Take a look around
and grab the RSS feed to stay updated. See you around!

The Declaration of IPv6 Independence

Somewhere in the 2020s, a decade after the last IPv4 address has been used up, computer science students are going to learn about the transition from IPv4 to IPv6 as an example of either the best or the worst system upgrade in the world's history of system upgrades. We know it will be the greatest system upgrade either way: hundreds of millions—if not more than a billion—systems will have had to be updated in some way. 老域名出售

The big question is: do we wait until a crisis erupts, or do we start addressing the problem now? According to the experts, we're rushing headlong into the crunch right this moment; ARIN says we have less than 20 percent of our IP blocks left, which means we could potentially exhaust IPv4 in just a few short years. Perhaps more importantly, IPv4 becomes more difficult to deal with and to manage as we get closer and closer to full capacity.

Admittedly, I'm a big proponent of IPv6. I've written a lengthy introduction to IPv6 here at Ars, and I've published a book on the topic, as well. Thus, my position should not be that surprising: let's start switching to IPv6 now, and let's see corporate IT lead the way.

The challenge is that the IETF has been pitching IPv6 as an addition to IPv4. This doesn't necessarily help things. Don't misunderstand me; the ability to run IPv4 and IPv6 side by side is an essential ingredient for a successful transition… for servers. But for everyday, non-server computers, dual stack isn't so great. In many ways, IPv6 is easier to set up than IPv4. With IPv4, automatic configuration via DHCP was an afterthought. It works well, but when it fails, it's very hard to fix the problem in heterogeneous, complex environments.

IPv6, on the other hand, was designed with "zero configuration" and a populated, busy LAN in mind; it pretty much always works. Apple even went so far as to add limited IPv6 capability to their WiFi base stations some years ago to make the installation and configuration process more robust. Plenty of other companies have done the same, and enterprise hardware is not always a slouch in this regard.

The octet sacrifice

No matter how easy it is to run IPv6, running IPv4+IPv6 is always going to be more complex than just running IPv4. Another downside: the IPv6 path to a remote destination is often longer than the IPv4 path, so performance (at first) can be lessened. For these reasons, we often hear that adding IPv6 to the mix is just a lose-lose proposition at this point in time.

The simple fact is that as long as you have IPv4, no matter how convoluted to configure or hard to obtain it is, it's less work than a complete migration. That's unfortunate, because sticking to IPv4 only means a future filled with more limitations, and hence more hacks to keep things running, while a collective move to IPv6 would fix everything once and for all.

So how do we break the vicious cycle? Simple: start planning now to turn off IPv4 as soon as you can. But wait a minute… how do I do my Google searches, read all those great Ars articles, or send the latest LOLCAT links to my friends? Won't client computers suffer? The short answer is maybe. The long answer is that IPv6 can get you where you need to be, and if not, a translation box (such as an HTTP proxy) can translate from IPv6 to IPv4 on your external network. For home users, this is too much trouble for right now. For medium and large businesses, as well as enterprise environments, the move could save you a lot of trouble in the long run.

In many corporate networks, network management is a serious headache, and firewalls break much more than 5% of all applications. IPv6 should be a savior in this regard, and don't forget: the applications that fail to work through the IPv6/IPv4 translation aren't lost forever; they just need to be upgraded to IPv6. You're going to have to do it sooner or later, and if ARIN and others are to be believed, the emphasis is on sooner.

So let's start looking at removing IPv4, rather than just focus on adding IPv6. Who's with me?


Both comments and pings are currently closed.

Comments are closed.