Posted by: Dave | February 24, 2010

No, the sky isn’t falling with IPv4

What’s this IPv6 stuff about?

Well, without going too far into the technical details, everything on the internet is served from some address on the internet. These aren’t the internet addresses you may be familiar with (e.g. https://codergeek.wordpress.com) but rather numerical addresses like 72.233.2.59. When it comes to the infrastructure of the internet, these numerical addresses what infrastructure experts work with.

Over time, the internet has expanded from a few hundred, then to tens of thousands and now billions of possible addresses. Each time, a new version of the internet protocol (IP) needed to be released to accommodate more addresses. The current version of IP that we use is version 4, abbreviated as IPv4.

After IPv4 became widely used, an experimental IP version 5 was tested. However, the geeks involved with this basically halted work and thought: hey, this whole thing of coming out with a new IP version every few years just because we run out of addresses is getting repetitive. What is needed is the one addressing scheme to rule them all! *queue Lord of the Rings music*

So what if we had 1 possible address for approximately every 2 atoms in the known universe. This should be more than enough for the future of humanity! And so IPv5 was tossed out, and IPv6 was born.

This is great. In the late 20th century IPv6 began becoming implemented and very geeky folks started actually using it. In fact, they said that by 2002 we would be out of addresses in IPv4 and would be forced to switch to IPv6!

It’s 2010 and we’re still on IPv4. What happened? Well, different folks have different concepts of “running out.” The average person thinking that we’re running out of addresses means all possible addresses are in use and we’re about to no longer have more addresses to use. The folks pushing for IPv6 define “running out” as no more available ALLOCATIONS of addresses. What’s that even mean?

Analogy Time:

Let’s say I invited you and 10 of my friends over to my house. You all show up and see a dozen donuts in my kitchen. I give you all a donut, and I grab one for my self. A normal person would think this is cool, everyone got a donut – awesome! A person pushing for IPv6 would immediately freak out as soon as I gave the last donut away screaming about how there’s no donuts left in the house! We’d all just stare at him like he’s an idiot, since we all have uneaten donuts still in our hands, including the crazy IPv6 guy.

It’s all about allocation vs. actual use!

And that’s where the controversy of IPv6 begins. A study last year showed that 95% of all addresses on IPv4 remain unused: http://www.isi.edu/ant/address/. At the same time, we’re going to run out of IPv4 allocations before December 21, 2012: http://www.potaroo.net/tools/ipv4/index.html. Funny thing, this is not contradictory. Think about it, one study focuses on actual address use (donut consumption) and the other one focuses on allocations (how many donuts are left for me to give out). Organizations in charge of allocations are screaming that we should move to IPv6.

Think about that for a second, we’re out of allocations but we’ve barely used the possible addresses. Why wouldn’t they just draw more attention to the fact we have many unused addresses? Well, that would mean that organizations trusted to efficiently allocate those addresses have failed miserably. Why would these organizations admit their own failure at such a critical task?

Funny thing is, we’ve been told IP addresses would run out in 2002, 2004, 2006, 2007, 2008, 2009 and now 2011. Why the inaccuracy? It seems all these countdowns for IP allocations fail to account for new IP allocations becoming available. How can that happen, you may ask? Simple, when internet companies merge, those in charge of IP allocations force the merged company to give up all of their IP addresses in favor of a new, potentially more efficient, allocation. This means more IPs are available to allocate.

The problem is, IPv6 is not a panacea for this issue. Having 1 address for every 2 atoms in existence is worthless if those addresses are as inefficiently allocated on IPv6 as they were on IPv4! How dare we even consider such blasphemy, IPv6 is going to cure everything, even cancer! /sarcasm.

Sure, IPv6 has some awesome features built-in like multicasting, which could reduce network traffic in the long-term. I have no issue with moving to IPv6 for those reasons. However, to argue we need to all convert to IPv6 because the IPv4 sky is falling – well, that just makes no sense. Those allocating addresses are just as likely to be wasteful with IPv6 as they were with IPv4. Then, what are we going to do?

Oh, and since I get this question a lot, an IPv6 address looks like:

1050:0000:0000:0000:0005:0600:300c:326b

Instead of using base 10 digits (0-9) to represent the underlying binary that is an IP address, IPv6 uses base 16 digits (0-F). This allows us to represent 255 as FF, thus using fewer digits. The fewer digits is more useful as these addresses are longer. Why base 16? Well, geeks are already accustomed to base 16. Anyone who has done even basic HTML, even on a MySpace profile, probably knows this from the way colors are coded.

However, just like our current IP system, you shouldn’t need to memorize those addresses. Instead, you’ll just go to websites as usual like https://codergeek.wordpress.com and everything will automatically be handled for you.

Advertisements

Responses

  1. This is true, at least for North America, but in other regions, where networks are still being deployed, there is a requirement of those inexistent donuts to address their networks.

    The unavailability of IPv4 addresses will force them to use NAT or other ip-sharing techniques that will either make them use more bandwidth (in case of peer-to-peer apps) due to requiring a proxy somewhere else to enable a communications channel.

    At least with IPv6, all the regions are starting at the same time, and there are now Global Policies for distribution of addresses, preventing cases such as the unused 95%, which would be equivalent of assigning 3 donuts to the first two guests and leaving the rest for the ones that come in later to deal with a High demand and a Low offer.

    It is worth noticing that we should not be ‘alarmed’ by the fact that IPv4 addresses are running out. It will come first for Regional Registries, and much-much later for end-users.

    But the good news is that IPv6 is now a reality, so we now have a working solution for this problem.

    • True, very many lessons were learned on IPv4. For example, I think moving to CIDR allocation instead of the Class A-B-C system was a significant improvement with efficient allocation of IPs. Also, moving to a system that doesn’t favor my particular country with regards to IP addressing seems logical, especially given the always changing demographics of the Internet. I think the biggest failure of IPv4 predates the RIRS and is exactly what you mention: the early adopters got significant IP allocations which currently are going largely unused.

      However, the fact that individual hosting companies are being granted /48s seems a bit excessive, seemingly on par with the scenario of Class A blocks of IPv4 addresses being issued. While this is a CIDR-style allocation, it’s like we’re back to granting class A’s again (way more IPs than anyone can reasonably use in the next few years). If IPv6 is to be the last time we implement a new addressing scheme, the RIRs folks overseeing IP allocations are going to need to be more conservative.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

%d bloggers like this: