IPv6 in Real Life

Hello! My name is Flameeyes and I have strong opinions about IPv6.

In particular, I've been arguing with many people over the years that, despite increasing adoption from the point of view of providers, there's very little interest from the point of view of end users, since outside of big tech, most of the websites people need to visit for their "daily life" do not support IPv6 at all.

Full disclosure here: both my current and previous employers are interested in IPv6 enough to have public trackers on adoption, and this project is totally unrelated to both, measuring a completely different metric and it is my personal opinion, not reflecting my employer's.

Metric and Measurement

Finding IPv6 trackers for ISP adoption is easy (see above), and it's also very easy to find trackers that look for "Alexa Top XX" domains. Both of these types of tracking are easy to execute in bulk, because they don't need manual curation at all. On the other hand, I would argue they provide a much less useful picture when it comes to "daily usage".

What I'm focusing on for this project, is to gather not just domains and sites, but whole services, and collect them by country. This has the unfortunate effect of requiring a lot more manual work, and a lot of country specific knowledge, which is why you'll find the selection fairly biased towards the countries I have first hand knowledge of.

The reason why I'm gathering services together, is because it's not uncommon for many services to use different hostnames and hosting providers between their "showroom" website and the site that includes user management. This is to be expected as, for many years now, we have been using hostnames (or domain names in some other cases) as security boundaries. So for instance, many banks would have one domain to show off their current account offers, which would be including scripts for analytics and similar, while a completely different domain would be used for online banking, which wouldn't be loading any of that.

Being able to access a bank's website but not their online banking shouldn't count as the bank having a valid IPv6 setup. Being able to see the supermarket's offers but being unable to place an order shouldn't count as the supermarket having a valid IPv6 setup.

Glaring Omission: Mobile Apps

Unfortunately, there is one very glaring omission from my methodology: I have no way, at the time of writing at least, to confirm which hosts are required for each of the companies' mobile apps. And this is actually quite important.

One of the common refrain I hear from those who believe IPv4 is already an entirely legacy technology is that Apple has been forcing app developers to ensure their apps can run on a phone that does not have an IPv4 address at all. Now, this does not mean that the server side needs to be reachable without IPv4, since DNS64 and NAT64 are acceptable ways to allow connectivity to IPv4 addresses from IPv6-only devices. But it might very well be that companies that didn't provide an IPv6-native Web interface have implemented already an IPv6-native mobile app backend.