You may have noticed that there has been a lot of noise about IPv6 recently, with the ‘World IPv6 Launch’ taking place on 6 June 2012. However, you might wonder why progress remains so low. Well, the main factor is that IPv6 and IPv4 are two completely separate protocols. IPv6 is not backwards-compatible with IPv4, meanwhile IPv6-only hosts are unable to connect to the traditional IPv4 Internet.
Unfortunately this distinction is a decision that was made over 15 years ago; but it is only now that we are faced with the issues this creates. It means that for a relatively long period of time we will have two separate internet ‘universes’ – one old (IPv4) and one new (IPv6).
The servers will run two protocols, thus being able to connect to both kinds of internet. In turn, each server will need to have at least two IP addresses – one IPv4 and one IPv6 – in order to be able to reach everyone. If your server has to initiate a connection, it checks which address family (meaning which ‘universe’, IPv4 or IPv6) the second party is connected to. If it is in only one of them, it uses the appropriate protocol; if the second party is dual stack, then IPv6 is chosen with fallback to IPv4 in case of any connectivity problems.
A commonly quoted estimate of the amount of time left during which we will have to run two protocols is 10 years. In the world of the Internet, this is a very long time – especially when you look back and see what it was a decade ago.
So what will happen to for instance your Internet access at home or on your mobile? Well, bearing in mind the millions of SOHO routers, access-points and other network devices installed at home, it poses quite a big problem.
To counter this, ISPs are working on large-scale NAT (network address translation) devices which can translate between IPv4 and IPv6 universes. This allows IPv4-only customers to access IPv6-only content. Address translation will provide the connectivity to the traditional IPv4 universe, but native IPv6 connections will always be better, faster and less prone to problems. These devices are really going to boost IPv6 deployment in the hosting industry!
We may also expect to see mobile phones with IPv6-only connectivity in the near future. They will first appear in Asia, possibly even this year.
At Leaseweb we’ve ensured we’re well prepared for the future; our network has been IPv6 ready and running dual stack for several years now, which means that we can now assign IPv6 to all customers. The advantage of our network infrastructure is that you don’t have to use a separate network card; Leaseweb runs IPv6 on the same infrastructure, switches and cables as IPv4, the only difference being the additional protocol used. Despite protocol incompatibilities, they can still safely share the same infrastructure without any interference. Furthermore, we don’t have separate bandwidth quotas for different protocols, we just measure traffic regardless of whether it is IPv4 or IPv6. The message is that despite the concerns around IPv4 and IPv6, if your hosting provider understands, and is suitably equipped to handle these concerns, the transition will be smooth.
Stefan de KoninkMay 26, 2012 at 0:51
One of the main thing I have been always puzzled about is that Leaseweb isn’t implementing stateless autoconfiguration for all its racks. It is obvious to see that a per switch allocation of a /64 subnet would work much more efficient for a customer (IPv6 without any configuration) then the current requirement of manually assigning addresses. Why isn’t Leaseweb just doing this can’t be the lack of addressspace ;)
A similar situation arrives for IPv4. Leaseweb is running DHCP, it has to, otherwise netboot would never work. Why not offer DHCP per default? Can’t hurt much that the customer doesn’t have to care about fixing its primary IP address.
Grzegorz Janoszka (Network Design Engineer)) • Post Author •May 30, 2012 at 13:30
Thanks for your reply, you bring up some interesting points.
Our goal is to make IPv6 as similar to IPv4 as possible. Our customers configure IPv4 addresses manually, that is why we ask them to configure IPv6 manually as well. We have already had some issues with rogue IPv6 router advertisements, that is another important reason for not using RA.
We use DHCP during the provisioning phase; all delivered servers have static IP configuration (the only exception is the new Cloud platform). This is because it’s what our customers expect. We provide the infrastructure, while management is done by our customers – they are experts and they expect flexibility which cannot be achieved with DHCP.
T35 HostingJune 20, 2012 at 13:00
Ipv6 is supposedly a much better version of IPv4 and therefore increasingly popular and easier to use. When hosting a site this is imperative.
Daniel WozniakJanuary 5, 2013 at 12:02
The fact that some *cough* ISPs are actually nearly out of IPv4 addresses will certianly speed things along.