Monthly Archives: June 2012

Why the Sky Isn’t Falling and IPv6 Might Not Be the Answer

The End is Near! Or is it?

From time to time we see the doomsday headlines in mainstream media about the “Internet-Armageddon” or whatever dramatic name the editor conjures up. It was supposed to happen in 2008, then in 2011 and now they’re talking about 2013. What I’m talking about is, of course, the point where all the four-part numeric IP-addresses will be allocated and the pool of available addresses is completely depleted.

As some of you know, computers on the Internet are identified by four numbers between 0 and 255. Typically, an address is written as for numbers with periods between them, for example 145.65.78.123. That means that there are (theoretically) 256*256*256*256=4,294,967,296 (about 4.2 billion) addresses. In real life, there’s a little less than that, since some address ranges are reserved for various purposes, but it’s still in the range of about four billion. So clearly, with a world population approaching seven billion, there aren’t enough addresses for everyone on the planet, especially when people in the industrialized world carry something like three Internet-capable devices with them at any given time.

You may also have heard that the adoption of the “fix” for the address problem, something called IPv6 (Internet Protocol Version 6), is running at a snail’s pace. Here in the United States, IPv6 typically makes up less than a percent of the total Internet traffic. So are we looking at a crisis similar to the US Federal Deficit problem where everyone is aware of the problem but no one does anything? No, at least in my opinion, the IP address problem isn’t even a crisis. In fact, it’s hardly a problem anymore, since the industry has already adapted itself to a world where there aren’t enough IP addresses to go around.

A Quick Summary on How It Works

An IPv4 (the “old” current system) address identifies computers using a four number scheme, for example 145.65.78.123. The first number can be seen as an “area code” and identifies the “network” of an address and the next number can be seen as a “sub area code” and so on. (In reality, each number can be subdivided into more than one “area code”, but that’s beside the point for this article). As a user, you rarely use IP addresses directly. Instead you’d type something www.cnn.com, which is then looked up in a global “phone book” called the DNS (Domain Name System) and translated into an IP address. So the textual web addresses you normally use are just a beautified version of the IP address that makes it easier to remember.

When you request IP addreses you can either get them one by one or purchase an entire chunk by claiming all the addresses in an area code. For example, IBM owns the entire 9.x.x.x area code. That’s almost 17 million addresses! A smaller company would request a smaller range, for example 145.65.78.x, which comes out to 256 addresses, minus a couple that are reserved for special purposes.

IPv6 (the new stuff), on the other hand, uses a whopping 16 numbers instead of 4. This allows for over 3.4×1038 adresses, a number so large that you could easily give every cell in every body on earth its own IP address without being close to exhausting the pool of addresses.

Why the World Hasn’t Ended Yet

In 2011, the final chunk of IP addresses was checked out from IANA, the international agency dealing with such things. So we already live in a post-IP-exhaustion era. How come email and web still works?

The answer is something called NAT (Network Address Translation). In simple terms, it’s a technique that allows several computers to share a single address. In fact, your home network most likely uses NAT. Your home has a single IP-address exposed to the Internet. This is normally the address of your modem or broadband router. Your computers are given “fake” addresses that aren’t visible outside of the network. This works because virtually all communication is initiated from your computers and not to it. When you ask for a web page, you initiate the session from your computer, so the fact that the address of individual devices isn’t visible to the world is not a big deal.

Even large corporations with hundreds of thousands of employees use the same NAT technique as your home network does. Your large bank may only expose a handful of “real” addresses, while all of its thousands of employees are happily typing away on computers with “fake” internal addresses.

The reason NAT provides relief against address space exhaustion is because the internal “fake” addresses are considered “junk” addresses that anyone can use without having to draw them from a world-wide pool. So if your huge mega-company uses NAT (which almost all of them do), they can get by with very few addresses, even if they employ hundreds of thousands.

The Problem with NAT

IPv6 activists don’t like NAT. In fact, their favorite sound-bite is “NAT doesn’t solve the problem”. They are only partially right. IPv6 pundits usually have two issues with NAT:

  1. It’s an ugly solution. NAT is a bit of a “hack”. A quick an dirty fix that definitely lacks elegance. But it works. In fact, when you read this, you’re probably doing so over at least one level of NAT.
  2. NAT impairs the point-to-point capabilities of the Internet. In the early days when everything on the Internet had its own address, any two devices could establish a two-way connection and it didn’t really matter who initiated the connection. With NAT, internal computers inside a home or corporation aren’t visible to the outside world, since they don’t have a real address.

The first complaint is easy to dismiss, since it’s of interest only to the most hard-core Internet protocol geeks. The second argument is more interesting, but mostly falls apart on today’s Internet. Today, very little communication is truly point-to-point. We typically interact with some central resource or go through a middleman. Also, from a security point of view, it’s obviously desirable to make internal computers inaccessible from the outside world. It’s important to know that NAT alone doesn’t replace a firewall, but it definitely provides an extra level of security. The only serious problem is with programs, such as certain games and file sharing software that require point-to-point communication. This can usually be handled by re-configuring routers, but requires a bit of skill.

The Problem with IPv6

So if NAT doesn’t provide a perfect solution, why hasn’t the world already moved to IPv6 if the new protocol solves all the problems of address exhaustion without the disadvantages of NAT? There are several reasons:

  1. IPv6 is not just a little tweak to the current Internet protocol. It’s a brand new protocol. That means that all your devices, computers and software must be upgraded to support it. Most modern computers and programs support it, but the problem is that if you ask any given organization what percentage of their IT assets support IPv6, they have no idea. This introduces an enormous risk!
  2. Since IPv6 is a new protocol, IPv6-enabled computers can only talk to IPv6 websites and other services. Today, about 25% of websites supports IPv6, which means that if you were to switch to IPv6 today, 75% of the Internet would become useless to you. There are ways around this, but they all involve somehow mixing IPv4 (the old protocol) and IPv6, which means more stuff to install and maintain.
  3. There’s a steep learning curve for IT professionals who want to deploy IPv6 in their organizations. Among other things, an IPv4 address, such as 192.168.1.12 can easily be given to someone over the phone or jotted down on a piece of paper. An IPv6 address, which looks something like this 2001:db8:85a3:8d3:1319:8a2e:370:7348”, isn’t as easy to communicate.
  4. As long as IPv4 still works, there’s going to be little financial incentive to move to IPv6. This may sound short-sighted, but from an internal IT perspective, large organizations have very little to gain from moving to IPv6, at least not internally. Because of this, there is no momentum to make the move.

Why the Internet World Won’t End Soon

I hope you’ve been enjoying our discussion so far and maybe even picked up some new knowledge. But the question remains unanswered: What will happen to the Internet over the next years and decades? In my view, it’s going to be business as usual to a large extent and the adoption of IPv6 will remain slow for the next 5-10 years, if not longer.

The reason for that is very simple: Capitalism. Let me explain.

While most people are perfect happy with NAT, some people with applications requiring point-to-point communication will still need dedicated “real” IP addresses. These people would have to pay some kind of premium to do this. Let’s say it costs you $5/month extra on your Internet bill to have your own IP address. While this isn’t a lot of money to most people, it has some interesting implications when you start thinking in terms of the monetary value of an IP address. If you can charge $5/month, it means that an IP address would bring you $60/year, so owning and renting an address to someone could be an interesting investment. Let’s say the market says an address, based on it’s future cash flows, is worth $100 (definitely an investment I would consider!)

Now, let’s talk about the A-networks! In the early days of the Internet, some large organizations, such as GE, HP and IBM were allocated huge chunks of addresses called A-networks. Each A-network has almost 17 million addresses. Now, consider that thanks to NAT, even a large company can get away with a lot less than that, maybe a couple of hundreds or thousands. If you’re the lucky owner of an A-network, maybe in some not too distant future, you can sell chunks of your block of addresses. At $100 a pop, you’re looking at 1.7 billion dollars for an A-network! $100/address is a bit arbitrary, of course, but even at a much lower price, selling your huge block of IP addresses you don’t use makes a lot of financial sense. This would then create a market for IP addresses, similar to stock and bond markets.

Although we might not have a market for IP addresses anytime soon, it is safe to assume that having your own address will come at a price, so people who don’t use them will want to relinquish them back to the pool again. People who don’t need point-to-point access will get a discount and placed behind a huge NAT with maybe a whole town sharing a single “real” address.

I personally believe we could go on for at least a decade without anything significantly bad happening even if the World doesn’t switch to IPv6.

Is IPv6 Dead In the Water?

Some people are starting to question whether IPv6 will ever catch on. I think it will, but very slowly. This is what I think will happen:

While most people would happily continue to use IPv4, there will be applications and devices that can benefit from IPv6. I think especially some mobile applications would benefit from point-to-point capabilities. So we will see an up-tick in IPv6 adoption in some markets. The rest of the world will muddle along with IPv4 without noticing much change.

To summarize, there’s no cause for alarm. You, as a regular Internet user will not be affected in any meaningful way anytime soon. If you absolutely need your own IP address, you may have to pay a fee, but that system should be fairly self regulating, since it would immediately refill the pool from people who don’t need and don’t want to pay for an address. This is why it bothers me when Internet pundits are claiming the sky is falling and media occasionally picks it up and blows it even further out of proportion.

Tagged , , ,