IPv6, where are we?
While chatting about the fediverse I was asked to comment "because IPv6 is not spreading", and at first this statement left me perplexed. I work on access networks, both mobile and fixed, and this is definitely not my experience. But beware, since my experience is not statistically relevant, I have decided to resort to a company that can give me an overview of the internet.
Obviously there aren't many companies like this, and luckily Google shares its data with the public.
As you can see, IPv6 is being adopted, its adoption is growing, it never happens to decrease over the years, and over the years it has reached 40%.
So my personal perception is correct: IPv6 is being implemented and spreading.
Obviously, a slice of all this is related to the mobile world, so many of the addresses that google reports on IPv6 are cellular devices, for which it is difficult to disable IPv6 completely.
Now some questions remain:
- How come it's not talked about?
- Are there any technical problems?
- Could it be faster?
- Is there something holding back IPv6?
let's try to answer.
Little is said about it because it is a slow migration, which today we would define as "canary": it means that there is no "golive day" in which you send production and poof, starting today all over the world we go with IPv6.
Journalism today has become hypercompetitive, and clickbait dominates. A title like "IPv6 arrives, how to get compliant in time otherwise your sister will turn to porn" is epic, it attracts clicks, produces anxiety because there is a date, and so on. Everything that happens when in Italy you have to change "er decodder".
A story like “IPv6 has passed 40% worldwide adoption” automatically ends up in the “meh” section of the paper, or on page 4 of Google. Who cares?
So little is said about it: after all, a newsletter that tries to inform people on the Internet cannot be called "News from the Internet", but must be called "Network Wars": even if there are no real wars on the net, it is enough anxiety-provoking, that is, engaging, to take hold. But news like “IPv6 has reached 40% adoption” doesn't smack of war. Perhaps "IPv6 troops have conquered 40% of Internet territory" would make more headlines, but then the article would be boring.
Migration goes on anyway, it's just that it does so in a little epic and little anxiety-provoking way, so it's not talked about much.
Will there be events to be featured in the papers? Well, in a few years it could happen that a Netflix or a DAZN say "from day to day we only go to IPv6, due to the undeniable advantages that the protocol offers for streaming providers". In that case there would be panic among users, and perhaps it will end up in the newspapers.
But until then you won't see that happen.
There are no longer (or have been resolved) the technical problems. The big issues (multihomed routers, or switching vs routing, and others) have been resolved. The NAT mechanism has also been imitated, (ULA, NPTv6, nat66(*), etc), so we can say that the main technical problems have been solved.
After all, it would certainly not be possible to reach 40% of IPv6 clients with an incomplete or faulty protocol.
When you hear someone say that IPv6 "has unsolved technical issues," you can safely tell them that they're incompetent and shouldn't get their hands on computers and network devices.
All right, then? Well, there are more critical areas. The M2M, Machine to Machine. There are many devices out there that have the IPv4 stack embedded , ie made on silicon, or in some ROM/EPROM/etc, which is not easy to replace. It will therefore be necessary to wait for their replacement for simple end-of-life, and years will pass.
And that's what happens in the mobile world, where the adoption of IPv6 grows as cell phones are replaced with new models. Here we are talking about ATMs, vending machines, gas, water, electricity meters, etc. However, the cycle of this stuff is not infinite, so the trend will remain increasing, but not fast.
The famous "security problems" are actually non-existent: defending a host remains defending a host, tcp ports remain such, routing is still routing (although different, almost all operating systems show routing tables in homogeneous), etc. The problem is that you have to update yourself and learn the new tools. Which aren't all that different.
I would say that, apart from the legacy systems, we can say that the technical problems are laughable or overcome. Anyone who says otherwise is simply incompetent or hasn't kept up to date. (yes, even your cousin who works for XYZ).
Could the migration be faster?
Definitely yes. Here we still need data from Google:
If you look at the data by country, you discover that India is the absolute giant in terms of population. 65% of addresses are IPv6. It is not the tallest in the world, but the number of inhabitants is so gigantic that it certainly stands out. Even the USA, which has always been recalcitrant, is around 50%, and since the market is very rich, this too is a good engine.
But who slows down?
If we go to Europe, we immediately discover an incredible thing.
The French have the lead, with 72% of users having IPv6. Germany follows with around 65%. The other countries are struggling, but the rear are two: Italy (7%) and Spain (3%).
What does it mean? It means that if tomorrow Netflix decides to go IPv6, in France few users would have to change routers or ISPs, in Germany slightly more, but in Italy there would be panic.
This explains why the person who asked me why IPv6 doesn't spread was Italian, while my perception differs since I live and work in Germany. Moreover, in Germany that 35% of pure IPv4 is due, in a large percentage, to legacy industrial devices. (but I don't have exact data to show you).
So it happened that a person who sees 7% of the use of IPv6 asked "why nobody uses it" to a person who sees it in 65% of cases, and that he had to buy a different router to be able to use IPv4 at home, since the “normal” router used only IPv6 (then the ISP , Vodafone/Unitymedia, does a NAT64 to go out, when needed).
Hence my surprise.
For the rest, the question "could it be faster" takes into consideration three things:
- Average age of the devices. In India they started building on a green field, so they adopted new devices from the beginning. So they started from IPv6. So it's much simpler. There are countries, such as Germany or France or England, which have so much industrial legacy that they reach, in my opinion, the maximum 75%/80%, and then wait for the old devices to be replaced due to old age. Countries like the USA, today at 50%, will move at the speed decided by google and apple, who make cell phones.
- So far all ISPs are keeping a double stack , ie they support BOTH IPv4 and IPv6. It's a more complicated situation than managing just one of the two. Only one of the two, if desired, can be learned off the cuff, but solid networking skills are required to manage the double stack well. Hence the number of incompetents who say they have "tremendous security problems" with the double stack. This obviously keeps the incompetent in the market, because having the double stack in itself allows the dinosaurs to find an ecological market niche.
- GAFAM tolerance. Nearly all of the large hyperscalers are still based on IPv4, although migrations are ongoing. Windows still has double stack, big services still use double stack. Until the US market is ready, they are unlikely to switch entirely to IPv6. The day any of them decide to ditch the dual stack and switch to IPv6, obviously some countries will experience a shock they will remember. BUT they will move fast, because they will have to.
Do I see a light at the end of the tunnel? No, not seriously: in the sense that there is no tunnel. IPv6 is not backwards compatible to IPv4, so we are talking about migration things. Excluding the possibility of going "green/blue", for obvious legacy reasons, such a transformation takes time. So far, the pace has been quite good, considering the size of the phenomenon.
It could be faster if some European countries (Spain, Italy, but also Ireland and Scandinavian countries) decide to adapt to the European average. But Spain is dominated by Telefonica, which has just sold parts of the access network because it wasn't earning much, and in Italy, Telecom Italia Mobile's financial problems are holding the bank with the government. Therefore, internal problems in the governance of the telco world are causing the delay.
Is there anything else holding back IPv6?
YES, the human factor.
The IPv4 protocol can also be learned “off the cuff” at the Home/SOHO/SME level. Thus, there are millions of people working in SMBs and professionals who use IPv4 without having a clear idea of what they are doing. But it works.
Even IPv6 by itself, if desired, could be learned that way, provided you have no idea what you're doing, and work more or less off the cuff.
The trouble is that, however, to manage the double stack you need to know what you're talking about. Otherwise it will happen that the most popular question on the internet will become “how do I disable IPv6 on Linux/Windows/MacOS?”.
This is because people who know nothing about networks try to use VPN/Routing/Firewall/Tor/Darknet and don't understand why a packet originates on a given protocol and follows a certain route. My answer is simple: if you can't manage a double IPv4/IPv6 stack decently, stop with rete and go to Lambrusco. (every similarity with a famous Romagna proverb is intentional).
So we have, due to the simple human factor, entire countries where the company or home network expert does nothing but disable IPv6, simply because the expert is not really an expert and can't handle the double stacks. (let alone tunnel or nat64 ).
The biggest obstacle in the diffusion of IPv6 lies precisely in the human factor: a gigantic number of people who sell themselves as network or even security experts, who have learned IPv4 by dint of "two hundred and fifty-five", and since they cannot manage a double stack disable IPv6 because it "gives it security problems".
I'll say it again: if you can't handle a double stack and/or it seems chaotic and/or it gives you “security problems”, simply stop dealing with networks: it's not for you. Or take a class. Or buy some books and study them.
For the rest, my answer is: the migration to IPv6 proceeds, not everywhere at the same speed, and the main obstacle is the telcos without money, or the incompetence of the single operators. The slowness is due to the size of the challenge: those of you who have ever carried out major renumbering or network migration operations can imagine how simple it is to "migrate the Internet".
Of course, when the "BIG" countries (by population or GDP) are ready, switches will start to flip, and in the least ready countries there will be panic.
For the joy of the newspapers.
(*) is not a typo, nat64 is something else.