My favorite IPv6 tunnel broker Sixxs.net recently announced, tongue-in-cheek, that they would be accepting no further new registrations and will also be throttling existing users. Those still interested should preferably contact their Internet provider and demand native IPv6 of them. The backdrop to this is that a large number of Internet and hosting providers would persistently refuse to offer native IPv6. Instead, these providers issue recommendations to simply register an IPv6 tunnel. Sixxs themselves would now put paid to that.
This message reached me on April 1. Sixxs were very friendly and had informed all previously registered users of this decision, whilst simultaneously requesting a “Call your ISP for IPv6’” campaign – as reported at heise.de and elsewhere.
Those who still hadn’t quite caught on, in light of the message’s very dubious delivery date (which is in Germany the day where people fool one another, were if anything surely puzzled by the still available option of registration. How many users followed the invitation to “Call your ISP,” I cannot judge. It emerged from an available customer survey from a legacy provider, however, that in the last week a surge in demand for IPv6 had taken place.
Reason enough for me to invest a few minutes in a little research.
I initially began with gridscale and determined that we barely register 10% of our traffic over the IPv6 protocol.
Does this figure surprise me? Well, to be quite honest, I haven’t really thought this through. To me, the obvious was clear. The share of IPv6 traffic would lie somewhere between 0% and 100% – the closer to the borderline scenario, the more unlikely the surge. But is 10% a lot or just a little?
Since we at gridscale waive charging for IPv6 traffic, current traffic costs will presumably play no major role here. So there must be other reasons why 10% of our traffic is IPv6 – or why 90% is IPv4.
When we were planning gridscale, the question about IPv6 did not arise at all. We have been working for years with IPv6 and it is considered standard by us. But we still, even today, see an incredibly large number of providers having IPv6 on their roadmap for several years without any results. Those involved here are not only legacy hosting providers, but with companies of supposedly modern IaaS and PaaS services also proudly belong to this group.
A good example of how to do it differently would be a showcase company such as Hetzner Online GmbH.
But what in any case are the drivers behind IPv6? And what is the worldwide picture in distribution?
First to the reasons. I’ll omit the reference to scarcity of IPv4 addresses, since this can be irrelevant to the user. That is merely a problem for the ISP’s. But for a while now, heavyweights in various sectors have began strongly pushing IPv6. One prominent example to name would be Apple.
Since early 2016, Apple has requested of its app developers that all apps support the so-called dual stack. No support for IPv6 thus implies that the doorman at the iOS store will leave you standing outside in the cold.
Also, network operators such as Deutsche Telekom, for example, have already adapted their mobile network to dual stack. In 2015, Deutsche Telekom thus became the first German provider to universally deploy IPv6. Others followed and are still following.
But back now to my brief research. The best statistic I found was from Google. It emerges here that about 10% of worldwide traffic is handled over the IPv6 protocol. The corridor for the expansion of IPv6 is thus enormously broad within the respective countries. The following map shows some European countries and their respective degree of IPv6 distribution.
From a worldwide perspective that corridor is even broader still. Countries where the degree of Internet acceptance is extremely low, logically also have no relevant spread of IPv6. I was never able to find a higher distribution rate than that of Belgium.
But how long will it still take before IPv6 becomes the quasi-standard? A potential answer to this is contained somewhat obscured in the Google statistic link.
Stichtag Verbreitung weltweit in % Multiplikator 01.01.2010 0,11% 01.01.2011 0,19% 1,73 01.01.2012 0,42% 2,21 01.01.2013 1,06% 2,52 01.01.2014 2,78% 2,62 01.01.2015 5,82% 2,09 01.01.2016 10,40% 1,79 01.01.2017 21% 2,00 01.01.2018 42% 2,00 01.01.2019 85% 2,00
If we assume a similar future growth based on the past, then we will all be speaking IPv6 in three years.
Be that as it may, gridscale is optimally prepared for this future. If you’re thinking to yourself whilst reading this article, “Hey, I know a particularly good source in this area,” then I’d be very grateful for a short message sent in to us. I will gladly link further useful sources about IPv6 in this article.