We have heard for years that IPv4 is over and that IPv6 is the future, but nothing has changed yet (or so it seems)
IPv4 addresses were out of stock years ago, and the world keeps turning. That seems to be the conclusion of a situation that some analysts called worrying and that years later does not seem to have affected the day-to-day lives of millions of users and companies around the world.
In fact, there was a time when various organizations alerted us to the situation and warned that we should prepare for the change to the new IPv6 protocol and its 128-bit addresses. That didn’t help much, and today IPv6 address support is barely over 25%, and we may never see a full migration to this IP address space.
Beware that IPv4 ends (or not)
Each device that connects to the internet has an IP address, a kind of DNIe or car license plate that identifies it and differentiates it from the rest. When the network of networks was designed, a protocol was thought of that would serve to assign these addresses in an organized way and with enough room for maneuver so that millions of devices could connect to the Internet.
Thus came IPv4, a protocol with 32-bit addresses that allowed the generation of IP addresses for 2 ^ 32 devices. Or what is the same, 4,300 million of them (specifically, 4,294,967,296). It seemed like enough room for maneuver, right?
How low-level protocols are changing the Internet
It was not. The arrival of smartphones and connected devices changed everything, and this address space was found to be insufficient at the beginning of this decade. New ID cards had to be generated for future devices, but there were no longer numbers available to do so. We had run out of license plate numbers.
There was a potential solution. The IPv6 protocol, which became a draft of an IETF (Internet Engineering Task Force) standard in 1998, would be ratified as an internet standard in 2017. Curiously, we had skipped the IPv5 protocol, which had a different purpose and became the so-called Internet Stream Protocol and did not go beyond the experimental field.
That IPv6 solved the problem with that 128-bit address space: 2 ^ 128 addresses were suddenly available for 340 sextillion devices (difficult to imagine the number, but believe us, it is more than enough for our short, medium, and even long term) had their particular identifier. There were license plate numbers to spare again.
The problem with IPv6 is that it was designed without backward compatibility with IPv4. Either you used one, or you used the other. Or, as it ended up happening, you used both at the same time to migrate from one to the other when IPv4 was finally redundant.
The problem is that this has not happened and could be far from happening.
IPv6 adoption slows down and may never reach 100%. The alert due to the lack of IPv4 addresses caused that, as we mentioned from various organizations, end-users and especially companies and internet providers – who had to enable this change – were encouraged to adopt IPv6 in all types of devices.
The adoption of IPv6 monitored by Google was 25.19% on August 7, 2019. According to Ericsson Research, the pace is not promising, and at this rate we may not even exceed 30%. That message did not seem to penetrate too deeply, not in users or in companies: the pace of adoption has been good for much of this decade, but the predictive models seem to make it clear that this growth does not point to a total migration to IPv6 or short term nor, attention, in the distant future.
A recent study by Ericsson Research made that clear. This company has been studying the adoption of the IPv6 protocol since 2013. It has been generating various models to try to predict the future state of this adoption and if at some point we were going to be able to talk about the definitive goodbye to IPv4.
Based on data such as those provided by Google – they have an updated graph that shows how many users use their search engine from a computer with IPv6 enabled – those responsible for Ericcson have been adapting their model, and the news is not good.
That final model has gone through phases. During the first years, the adoption was very fast, and in fact, it was thought that the growth of the adoption would be exponential. Ericsson engineers came up with models that indicated that by 2022 adoption would be around 80%.
However, at the beginning of 2018, they noticed that the forecasts were beginning to deviate too much from the real data, so they abandoned those previous models to adjust them and create the current model, which is much more in line with the real data.
Those optimistic predictions have been contrasted in the last two years with a much darker outlook for IPv6. This protocol does not appear to be being pushed or adopted by large corporations and agencies that run the internet.
Another added difficulty is that although IPv4 addresses have been exhausted in theory, they have not done so in practice. For years, various “IP brokers” have emerged, resellers of IPv4 address packages that are no longer used for one reason or another. These intermediaries sell to companies and organizations that need them for their users. Yes, the owners can monetize IPv4. One of them is through IPXO. It is a platform that facilitates IP brokers and those who need IP addresses.
This has made that “overwhelmed” by the lack of IPv4 addresses is not as problematic as it might seem. In turn, that raises more doubts about the need to adopt an IPv6 that, for now, is good to have as an alternative, but that does not seem to be imposed.
Not in the short term, of course. And maybe never.
Mohit Tater 2021-01-20 14:08:28