Running out of Internet

When I read the heading for this I almost laughed, but decided to read the article to see what it actually said. (No one would REALLY tweak the headline of summary to generate hits, would they?)

The allegation in the article is that 85% of all IP addresses have been used, and the remaining 15% will be used within 3 years. First the gas crisis, now an Internet address crisis! How much more can we handle?! And how true is it?

Let’s see if the claim makes any sense what so ever.

How many potential Internet addresses are there? If you have keyed in an IP address, you know that you enter four numbers separated by periods, or that you enter a number into each of four boxes. The valid range of number is 0 -255, for a total of 256 numbers. So there are a total of 4,294,967,296 possible addresses.

However, some numbers have been reserved, reducing the total available numbers for public consumption. Familiar ones include:

• 127.0.0.0 – “Home”, or what you can use to tell your computer to talk to itself. This is only 1 address out of the total possible.
• 10.10.x.x and 192.168.x.x are reserved for private networks. If you are behind a router at home, you will probably get one of those two group if you do an ifconfig or ipconfig. (ifconfig for PCs, ifconfig for Linux machines, which includes Mac users.) This eliminates another 131,072 addresses (256*256 or 65,536 each for the 10.10 series and the 192.168 series).

In addition, there are some administrative addresses, but the total available addresses is still in excess of 4,294,500,000. Though that seems like a lot, the current world population tops 6 billion, so that is less than one per person, and when you start considering all the equipment that is IP-addressable, things such as routers, servers (how many of those does Google have?), network-connected copiers and printers (we had six at my small school, and I have 2 at home), smart refrigerators that can let you know that you are out of milk and the meatloaf is 3 weeks old, unattended soda machines that call home to let the distributor know they need more Cokes, IP addressable traffic monitors, and so on, the claim suddenly starts making sense.

But does it really? How is the Internet organized? Let’s start with a global view. If you buy an IP address, you are buying one in your continental area. For me, that’s the USA, part of the North American set. Asia and Europe each have their own sets. So, as long as there are some interfaces between the two – and I can browse the BBC’s news service so I know I can get from America to Europe, and I talked via Internet phone to my son in Yap so I know I can get to the South Pacific – then we have effectively multiplied the number of IP addresses by the number of continental zones we set up. And if we need more, subdivide a “full” continental zone into several sub-zones and we have more. And there are maps that show connectivity and ownership of the entire North American Internet, while other maps show physical connectivity for specific owners.

The owners of the backbone provide access to their customers, major regional backbones, through the same type of interface as they use to talk to each other or to their peers on other continents. Likewise, the regional backbones talk to their clients on the local backbones using the same interface, and the local backbones provide service to the local ISPs the same way.

The local IPS – and there are thousands of these – provide service to the local customer through various types of connectivity. Most of those customers will be homes or small businesses with no need for any type of activity beyond email and browsing. A thousand local ISPs could each assign the same IP address to one of these customers, and everything would still be perfectly fine. If you execute the trace IP command, you will see that a request for info from google.com will go from your machine through your router, your DSL modem, your ISP, the regional backbone, maybe the main backbone, a regional backbone, google, and then back again. Or may cross through several IP addresses in any of those segments. If you and I have the same IP address, the trace info gets back to us because it knows where to go because the address layers are built up like an onion, and on the way back through the layers of onion are stripped off until the information packet gets back to our ISP, and then it knows that the address it is dealing with is inside it’s sphere of control, and hands it to your DSL modem, which hands it to your router, which hands it to your computer instead of your parent’s computer.

The number of IP address can become an issue when you request a domain name. “MySmallBusiness.com”, for example, which does not exist at the time I wrote this, might be assigned address 101.102.103.104 (an my apologies to whomever this might actually belong to). Since that is good anywhere in the North American IP group as opposed to being within a specific ISP, it does actually eat away at the total number of IP addresses available – the 4,294,500,000 or so mentioned above, and reduces the 15% available by a minute fraction.

So, the concern that we are running out of Internet is true, it is a real concern, but for businesses who what a unique IP address/domain name on the Internet. For home users, like me writing this, it is not an immediate or major concern.

Regardless, how can the problem be addressed? Fortunately, it has already been addressed. IPv6 is being phased in throughout the support infrastructure – at least here in the States. It supports 2 to the 128th (sorry, I can’t do superscripts/exponents in WordPress, but that’s a 128 bit address field) IP addresses, compared to the 2 to the 32 (a 32 bit address field) IP addresses currently available in IPv4. (They are skipping IPv5.) That’s a huge difference! And once that’s all in place, this IP address shortage crisis will be in the past.

Some great information about this can also be found in Steve Gibson’s and Leo Laporte’s Security Now Episode 25, and look about half way down. They also explain why the interfaces between the various continental backbones, the regional backbones, the local ISPs, etc., allow the multitude of addressing schemes to work now, why we can have multiple people (all on separate ISPs!) with the same IP address – the hint is “routers”, but I’ll leave the magic to them.