IPv6 was standardised around 15 years ago, primarily as a way to address the anticipated exhaustion of IPv4 address space, and since then we have had a steady trickle of people asking whether they could have one to use on the University’s network. Until now we have had to turn these down, unfortunately, as our network kit did not have enough in the way of facilities to allow us to support this properly.
Now that IPv4 addresses have finally been used up, we expect that sooner or later major ISPs will start to issue IPv6 addresses to their customers. Many large service providers, such as google and amazon are already fully IPv6-ready, for this reason. It’s inevitable that we will need to speak IPv6 in order not to be cut off from what can only be a growing part of the Internet.
With the recent upgrade of many of the Forum edge switches, and the intention to upgrade most of the remainder over the next year or two, we now think that we have enough support in place that we can start to take a proper look at implementing IPv6 locally. To that end, we have begun a Development Project to undertake an initial investigation into what will be required to support IPv6 within Informatics. It is not expected that we will stop using our existing IPv4 allocation any time soon, if at all, however. Indeed, it’s likely that IPv4 and IPv6 will run in parallel for quite a few years yet. The ultimate goal is that any Informatics system which needs an IPv6 address will be able to have one, or even many of them.
Please note that adding IPv6 support to our network is not something that can happen overnight. On the contrary, it will require that we turn things on in a gradual, phased way, testing carefully at each stage before moving on. If we do happen to miss something, there is the potential to cause quite widespread problems for existing IPv4 users, so needless to say we will be proceeding with caution! The expectation is that there will be several more projects requiring to be spawned by this one before all of the necessary facilities are in place. Managed machines are likely to be supported first, with self-managed machines coming later as a result of the extra infrastructure that will have to be in place for them.