A few weeks back, I was at a Microsoft TechNet UK event, where Steve Lamb discussed Microsoft’s implementation of the Internet Protocol v6 (IPv6), available in Windows 2000 service pack 3 or later, Windows XP service pack 1 or later, or Windows Server 2003. This is a new version of IP (also known as IP next generation – IPng), intended to overcome some of the limitations of the present version (v4), namely:

• Exhaustion of available addresses – not such a major issue now that network address translation (NAT) is so common, but potentially a future issue as more and more devices are IP-enabled.
• Large routing tables in backbone routers (the average ISP has 90,000 entries under IPv4).
• A need for simpler, stateless configuration.
• A need for better support of real-time data delivery (QoS)

IPv6 provides a 128-bit address space (compared with IPv4’s 32-bit implementation), and instead of being represented using four octets in dotted decimal notation, IPv6 addresses use eight groups of four hexadecimal digits, which incorporate the media access control (MAC) address of the client, for example, 21DA:00D3:0000:2F3B:02AA:00FF:FE28:9C5A

I’m told that there was an IPv5 (presumably with a 64-bit address space?), but it took too long to ratify. The IPv6 addressing scheme gives a vast number of possible combinations (about 340 undecillion – that’s more than 340000000000000000000000000000000000000!) and allows for faster routing due to its simplified header.

Like most protocols in the TCP/IP suite, IP is made up on an number of sub-protocols and IPv6 is actually formed of five core protocols:

• Internet protocol (IP).
• Internet control message protocol (ICMP).
• Multicast listener discovery (MLD).
• Neighbor discovery (ND).
• Top level aggregator (TLA).

(Yes, there really is a three latter acronym called TLA!)

In terms of application support, Microsoft’s IPv6 implementation is as per the IETF RFCs (i.e. not extended in any way). The tools look similar to the IPv4 versions, apart from the different addresses. DNS and RPC are both supported by the IPv6 stack, as are sockets interface extensions; however IPSec on IPv6 is only partly functional. There is also support for an IPv6 IP Helper API.

So what are the barriers to IPv6 adoption? For a start, businesses will need to see some benefit first, and although IPv6 addresses are available now, the initial worries about a lack of IPv4 addressing space have been alleviated (for the time being) with the use of network address translation (NAT) and private IP address ranges. Organisations implementing IPv6 do not need to drop IPv4 and convert overnight – it is possible to mix and match there is a world-wide IPv6 test network backbone; however, many organisations are using NAT as a line of defence in their security model and so firewall configurations will need to be re-examined if an IPv6 migration is performed. Add to that the fact that IPv4 is well understood by administration staff (IPv6 is not), a critical mass must build up before most organisations will be ready to make the move, although the the US government is mandating that all federal agencies must use IPv6 by 2008 – maybe that will start the ball rolling.

In summary, IPv6 is here today, but many organisations will not be in a rush to migrate. The next generation of Windows (codenamed Longhorn) is expected to include a new networking stack that supports both the IPv6 and IPv4 networking standards and I would expect IPv6 to gain some momentum around about the time of it’s expected release (2006-7). Until then, IPv6 will remain something to look at in our labs. Wikipedia has more information about IPv6 for those who wish to learn more.