IP Version 6

As pointed out earlier, the explosive growth of the Internet has pushed it to the very limits of the original design. The number of IP addresses has grown almost exponentially in recent years, leading to the fear that there will soon not be enough addresses to connect all the devices and computers that people want to connect. If that weren't bad enough, routers are being overloaded by the increasing inadequacy of the IPv4 standard.

To find a solution to the limitations of the 32-bit address space and the limits to the routing protocols in the current structure of TCP/IP, the IETF and others began working on a new version of IP several years ago. Different working groups originally proposed different solutions, but over time these groups have arrived at a consensus on the next generation of IP, IP version 6 (often shortened to IPv6). It was formally accepted by the IETF in December 1994, and the current specification is in RFC 2460.

IPv6 defines a 128-bit IP address space compatible with the current implementation of TCP/IP (version 4, or IPv4). The specifications call for packets to include additional information for improved routing and handling of mobile devices. IPv6 will not only enlarge the address space available, but it will also improve network performance, ease configuration issues, and provide enhanced security.

Real World

Unreal Numbers of Addresses

IPv6 addresses have four times the number of bits that IPv4 addresses do (128 vs. 32), but what does that actually mean? The 32-bit address structure can enumerate more than 4 billion hosts on as many as 16.7 million networks, but the number of potential IPv6 addresses totals 2⁹⁶ times the size of the IPv4 address space (4 billion ⁹⁶). This works out to 340,282,366,920,938,463,463,374,607,431,768,211,456 possible addresses. Of course, given routing and hierarchical requirements, this theoretical address space is diminished greatly when making practical estimates. The mathematically minded should consult RFC 1715, in which Christian Huitema analyzes other addressing schemes (including the French and U.S. telephone systems) and concludes that 128-bit addressing will suffice for another 25 years of Internet growth.

Despite the benefits, many are not looking forward to the deployment of IPv6. After all, most companies try to minimize the number of protocols they need to support. Supporting yet another protocol is regarded with less enthusiasm than Friday afternoon performance audits. The transition will be gradual, however, and there is an incredible amount of work in progress to ensure that it is painless as well.

Although the impending shortage of IP addresses has been pushed off somewhat by companies that use a small number of Internet-accessible IP addresses to provide Internet access to an internal private network (which is free to use a class A network), the time for IPv6 is approaching. Some companies, such as ISPs and cellular phone networks, are already deploying IPv6, although it will still be some time before IPv6 actually replaces IPv4.

Unfortunately, Windows 2000 doesn't ship with built-in support for IPv6. There are test-quality IPv6 implementations available for download on Microsoft's Web site (which operate side-by-side with the IPv4 stack), but these implementations are currently intended for software developers and testing purposes, not for actual production use. A fully supported IPv6 protocol stack might be available at some point for Windows 2000; otherwise, consider upgrading to Windows .NET Server when it becomes available, which is slated to have built-in support for IPv6.