What happens when we run out of IP addresses? by Steve Cunningham

8/23/2010 4:07:19 PM
By PSN Staff

When compared to the country's continuing economic problems, this month's technical issue looks like small potatoes. Still, the fact that we're running out of IP addresses will affect anyone involved in classroom technology applications. It's true -- we really are running out of IP addresses, at least while we're using our current IPv4 scheme. The only real solution to this problem is to fully implement IPv6, which turns out to be more complicated than it appears. A brief look at our current IP addressing scheme reveals the reasons why.

As most professionals know, IP addresses using the IPv4 scheme consist of a set of four dotted triplets of numbers, e.g. 123.456.789.012. This represents a 32-bit address, and yields a total number of about 4.3 billion unique IP addresses. That sounds like a nearly unlimited number of addresses until you consider the fact that every device with ethernet connectivity needs to have a unique IP address. That includes every network printer, PTZ camera, network-controllable projector, and so on. Suddenly a 32-bit IP addressing scheme seems merely adequate.

The fact is that this is a bullet we been dodging for some time; experts became concerned with depletion of the IP pool in the late 1980s as the Internet grew. That concern increased with the advent of always-on Internet connectivity in the 1990s. Yet there have been factors that have mitigated the problem in the face of ever-increasing number of personal computers, and peripherals in use worldwide. The widespread use of NAT routers has substantially reduced our need for more unique IP addresses over the past two decades. Because every ethernet device located behind a NAT router shares a single connection to its network's ISP, only that connection requires a single unique IP address. The devices behind the router can use a non-routable IP block like 192.168.xxx.xxx or 10.1.xxx.xxx to communicate with each other. This use of private networking has reduced the rate of depletion for many years. However, the widespread adoption of smart phones whose digital communications are IP-based has acted to again increase the rate of depletion.

The current consensus is that at the current rate of consumption we will run out of IPv4 addresses somewhere in the middle of 2011. The industry may yet be able to squeeze a few more addresses out of the IPV four scheme; a report done by Google in 2008 indicated that nearly 40% of IP addresses were not being used, either because they were non-routable like the 192.168 series, or because they belong to large companies and universities who got into the IP address game early on and claimed entire class A networks regions for themselves. A class A block consists of all IP addresses that begin with a common triplet (164.xxx.xxx.xxx), yielding 16 million unique IP addresses. Some companies that were initially granted class A blocks no longer exist, and these can be added to the available pool. Moreover, some universities have relinquished their blocks; notably, Stanford University relinquished their class A block in 2000.

Nevertheless, at some point we will run out. What happens then? The Internet Assigned Numbers Authority (IANA) believes that it will still be able to reclaim enough to IPv4 addresses to avert any calamity. But the only long term solution is to implement IPv6 across the Internet, and soon.

IPv6 addresses present themselves as four dotted quads of numbers, e.g. 1234.5678.9012.3456, which takes us from 32 bits to 128 bits, giving us on the order of 10^38 addresses [10-superscript 38], which is something like 340 trillion trillion trillion.

Computer operating systems have been IPv6 capable for a long time. Windows Server 2003 and Windows XP can do IPv6, as can Macintosh OSX and, for that matter, OS9. After all, the IPv6 spec was ratified in 1998. There is some level of deployment already in place on the so-called “6bone,” which is an IPv6 backbone. There are university environments which are using IPv6 experimentally. Still, the aforementioned Google report claimed that only 1% of the Internet was IPv6-ready. For example, our NAT routers at home, our Internet switches at work, and other small but necessary peripheral devices in our networks will likely have to be replaced. Keep in mind that switches and routers build a table of IP addresses, and memorize which IP is connected to which port. All of that happens in hardware, and all of that is currently done using IPv4, so it simply won't work. Many of the new features of IPv6, particularly those concerned with security and encryption, are not backward compatible with IPv4. So even though IPv4 addresses can be tucked into a corner of the much larger IPv6 address space, hardware or software translators may still be necessary.

When the move to IPv6 does happen, and it will happen in the near future, there will be obsolescence. In the meantime, we can watch and wait, and buy network-connected equipment that appears to be IPv6 ready, because we do in fact live in interesting times.

Share This Post