It’s been over ten years since the launch of IPv6, the Internet address system that was supposed to give us nearly infinite number of IP addresses. But after a decade of implementation, just over 21% of all websites worldwide will support IPv6 addresses.
On the surface, this makes no sense. The original IPv4 standard only supports about 4.3 billion IP addresses. We clearly have far more people and devices connected to the Internet than that number. By contrast, IPv6 provides 340 trillion trillion trillion IP addresses, a number that, for all practical purposes, is unlimited.
Even though we exhausted the supply of IPv4 addresses years ago, it doesn’t look like there is any rush for most of the world to move to the new IP addresses. There are obvious barriers to making the conversion that most ISPs and businesses are not ready to tackle. Most of the barriers to making the conversion can be categorized as hardware limitations, lack of training, and the overall cost of the conversion.
It’s a little hard to believe after a decade, but many older computers, servers, and routers will still not recognize IPv6 addresses. One would have to think that we’ll eventually ditch the older devices, but there are apparently still a huge number of devices that can’t process IPv6 addresses. The good news is that newer operating systems and devices will handle the new addresses. But the world still has plenty of folks using older versions of Windows, Linux, Android, and iOS. Big corporations are reluctant to make the switch to IPv6 out of fear of older technology around the company that would stop working. Smaller companies are not willing to make the change until they have no choice.
This issue is compounded by the fact that direct communication between IPv4 and IPv6 devices is impossible, and all exchange of data must pass through an IPv4/IPv6 dual-stack conversion to enable communications. This was originally envisioned as a temporary fix, but as IPv4 continues to be used, this is looking to be permanent.
Companies are also loath to tackle the cost and effort of the upgrade without some compelling reason to do so. Companies that have made the change report a number of unexpected problems with a conversion that can be disruptive, and companies are not willing to tackle something this complicated unless they have to.
It’s interesting to see how various countries have decided to make the switch to IPv6. Google has been collecting statistics on IPv6 conversions that are summarized on this map. At the time I wrote this blog, the world leaders in conversion to IPv6 are France (75%), India (68%), Germany (67%), Malaysia (62%), and Saudi Arabia (61%). Much of the rest of the world is far behind with the upgrade, including Russia (7%), China (3%), and much of Africa below 1%.
The US is just above 50% utilization of IPv6. Interestingly, the US backslid and was at a 56% IPv6 conversion rate in 2019. The resurgence of IPv4 is being credited to the huge flood of folks working at home during the pandemic – since residential ISPs have mostly not made the conversion.
Internet experts believe we’ll still be running dual IPv4 and IPv6 networks for at least a few more decades. We’ve found ways to work around the lack of IPv4 addresses, and very few companies or ISPs are seeing any urgency to rush toward a conversion. But as the worldwide penetration of broadband continues to grow and as we add more connected devices, the pressure will increase to eventually make the conversion. But don’t expect to see any headlines because it’s not happening any time soon.
Actually the original IPv6 specification (see the IETF’s RFC 2460) is over twenty years old
You’re right that the cost to migrate is an issue but overlooked one of the other reasons that many ISPs have been slow to move from IPv4, especially for consumer-focused services, which is the emergence of CGNAT. This has removed the urgency for many to migrate as it takes some of the pressure off that was being driven by IPv4 address exhaustion and significantly weakens the business case.
While it’s true that there’s less urgency for ISPs to upgrade their network to IPv6 with CGNAT available, they’re not really the crux of problem. That would be an easier fix. The issue is all of the IPv4 content out there. An ISP needs to have a publicly routed IPv4 address for every subscriber so that any customer can access any IPv4-only content that’s out there, even if the ISP is greenfield and starting out with an all IPv6 network. The cost to buy IPv4 addresses today are generally over $50 per address, so that’s an expensive proposition at the start, and the outlay continues as as the provider grows their business. We offer a CGNAT solution to ISPs for just that reason. They can oversubscribe one IPv4 address by up to 128:1, which saves them a tremendous amount of money. The IPv4 conundrum will remain an issue as long as there is IPv4-only content out there. It is expected to be that way for years to come.
I’ve had ip4 an ip6 on my phone for years. Both work fine
The US is at a 50% IPv6 conversion and cellular companies lead the way.
It is likely that the IPv4 address on your phone only exists between the apps and a 464XLAT pseudo interface on the phone – with many telcos, the path from phone to telco is IPv6 only. Telcos have realized gigantic savings by getting rid of IPv4 except within the phone.
I am currently involved in migrating one of the larger US Govt agencies from IPv4-only to dual stack then eventually to IPv6-only. There are many reasons for doing this (security, restoring end-to-end model, etc). There are also some mandates that REQUIRE them to migrate. The real savings and improvements come with IPv6-only. As long as there are any IPv4 addresses being managed and translated, we will have problems. CGN has been a major problem. Interpol has already called for it to be abandoned (almost impossible to track criminals). We are simply using IPv4 WAY past its shelf life.
I have been pioneering REAL P2P connections over IPv6, using PeerTLS (using a client cert at both ends for real mutual strong authentication and end-to-end encryption). Not possible with NAT in the way – easy with IPv6. With IPv4, to do “P2P” you have to use NAT Traversal, which is a major security problem.