Many things have been said about Windows 7 and its benefits. However, the talk rarely covered the topic of IPv6. Surely it isn’t a new feature: Vista was fully Ipv6 enabled and an IPv6 network stack exists for Windows XP. But in my opinion IPv6 is worth mentioning when talking about Windows 7.

Most networks today are still based on IPv4 because the Clients are not IPv6 capable. A lot of companies hesitated to migrate to Vista and still run Windows XP without the IPv6 stack. However, this circumstance will change with the release and deployment of Windows 7. Might IPv6, which technically is an old story, become more attention when Windows 7 is deployed en masse?

Not everybody is able to switch to Ipv6, though, because some needed applications may not support IPv6 yet. We have a hen-and-egg problem here: Most of the applications will only support IPv6 if most networks will run on IPv6, but the network administrators will only switch to IPv6 if all needed applications support it. Some argue that a killer application that only runs with IPv6 is needed to make people switch. But there is no such application available yet and the ones included with Windows 7 – Homegroups and Easy Connect, e.g. – can’t be called killer applications.

But maybe things lighten up a bit if we take a look at the improvements IPv6 has to offer over its predecessor:

  • Large Address Space: IP addresses in v4 are 32-bit long; in v6 they are 128-bit
  • Better Security: IPsec is part of the IPv6 standard, which makes the encryption of IP packets possible and guarantees their authenticity
  • Integrated Quality of Service: The IPv6 Header provides a Flow Label. It enables routers to make decisions and treat those packets in a special way
  • Mobility: With IPv6 a mobile device can be accessible by the same IP all the time

This short list only shows the major improvements. The listed points make IPv6 look like it does not have much to offer for lightening the daily work: IPsec, Quality of Service, and a static mobile IP are already possible with IPv4. They are not part of the protocol, but are in productive use and there is hardly any reason to change one’s running environment. Only the address shortage can cause administrators some headache, but it doesn’t affect Western countries that much. They have a big part of the pie and won’t run out of addresses in the next few years.

Especially in Africa the address shortage is a problem, though, and will become a huge one in a few years. It might even hinder the development of the continent’s IT infrastructure. With this in mind you could argue that we are morally obliged to migrate to IPv6 to support the development of third world countries. But usually there is no space in businesses for arguments of that kind.

But there is another good argument for migrating to IPv6: The address shortage already does cause problems for all internet users. Most administrators are not aware of them, though. The problems are rooted in the technique for overcoming this shortage: Network Address Translation (NAT).

Thinking that NATed IPs are an equivalent to public IPs is an error. Besides reducing performance, NAT does cause problems which are usually hard to discover. The reason for those problems is that NATed addresses break the end-to-end design specification of the IP protocol.

One of those problems is that the NATed address needs to be looked up in the Transition Table. Because the router is not informed if an IP Address of a client changes, it is hard to keep dynamically changing IP addresses up to date in the table. So it is quite common that entries of the Transition Table are out of date. This causes connectivity problems for video conferences, Voice over IP and slows down P2P downloads, etc.

Besides that the router is only able to maintain the Transition Table well for TCP connections. For stateless protocols like UDP the router has to guess if there will be more packets. Activity Timers are used to make the guesses more educated but they fail sometimes. The appropriate and valid entry from the Transition Table gets deleted then and as a result the connection dies.

NAT changes information in the header of IP packets. This causes problems with some protocols like IPsec, because it does integrity checks on the packet header. Another technique called NAT Traversal is needed as a work around. However, not only does it complicate things but it also is not standardized, which causes further problems.

This is by far not everything which needs to be said about IPv4 and the problems it causes today. A simple solution would be to use IPv6. So do you think that with the upcoming release of Windows 7 IPv6 will get more popular?