After decades of work toward adopting IPv6, APNIC Chief Scientist Geoff Huston has proposed that fully replacing IPv4 might no longer be essential.
IPv6 was designed to address IPv4’s limited supply of IP addresses, but due to advances in Network Address Translation (NAT) and Content Delivery Networks (CDNs), IPv4 has remained viable for much longer than expected. Huston’s analysis suggests that the urgency behind IPv6 adoption has faded, and he advocates a more flexible approach: “IPv6 was intended to be the future, but maybe the future is both IPv4 and IPv6 working together,” Huston argues.
Geoff Huston is the Chief Scientist at APNIC and a highly respected figure in internet development after bringing the internet to Australia in the 1980s and 1990s and through his extensive experience in internet architecture, protocols, and address management.
IPv4’s Address Problem and the Goal of IPv6
IPv6 first emerged over 25 years ago to solve IPv4’s limitation of roughly 4.3 billion unique IP addresses, a limit that quickly became insufficient as the internet expanded globally. Designed with a 128-bit addressing system, IPv6 theoretically offers trillions of unique IP addresses, aiming to make “address scarcity” a thing of the past. Yet, adoption has remained modest, with only about 40% of internet users currently using IPv6. Huston observes, “The IPv4 address pool was officially exhausted over ten years ago, yet we’re still managing to meet demand without a full transition.”
The Global Use of IPv6 in % (APNIC)
According to APNIC, which manages IP address allocation in the Asia-Pacific, the limited adoption of IPv6 can be traced back to high implementation costs and minimal perceived benefits. Instead, many providers found ways to extend IPv4’s capacity using alternatives like Carrier Grade NAT (CG-NAT) and CDNs.
Current data on IPv6 adoption by region from APNIC reveals stark variations in IPv6 capability and preference across global subregions. Southern Asia leads with the highest IPv6 capability rate at 71.38%, followed closely by Western Europe (62.69%) and Northern America (53.63%). In contrast, regions such as Micronesia and Southern Africa show minimal IPv6 capability, with rates below 2%. Interestingly, despite high capability rates, the actual IPv6 preference remains slightly lower in most regions, indicating that while IPv6 is available, it’s not always the preferred protocol for users.
Weighted sample counts emphasize the high testing volumes in regions like Eastern Asia, where nearly 190 million samples reveal a 42.07% IPv6 capability. This disparity in IPv6 adoption highlights ongoing regional gaps in the global transition away from IPv4, underscoring APNIC Chief Scientist Geoff Huston’s view that IPv6’s widespread adoption is facing considerable challenges and may continue to coexist with IPv4 in many regions.
NAT and CDNs as IPv4’s Lifeline
One of the primary reasons IPv6 adoption has stalled is the development of NAT, which allows multiple devices on a local network to share a single public IP address. “NAT has been nothing short of a lifesaver,” Huston explains. “It’s allowed IPv4 to keep up with demand in ways we didn’t predict.” NAT operates by concealing individual device addresses behind one public IP, which has made it essential for internet service providers (ISPs) needing to support thousands of devices without IPv6.
CDNs—networks of servers distributed globally that cache website content closer to users—also reduce strain on IP address resources. With CDNs, domain names and DNS (Domain Name System) direct traffic to local servers, allowing users to access content faster. This reliance on names over numbers has made IP addresses themselves less central to routing decisions. “CDNs prioritize names over numbers,” Huston points out, noting, “The entire model of traffic delivery shifted, making unique IP addresses less relevant for end users.”
“We’ve been able to take a 1980s address-based architecture and scale it more than a billion-fold by using NAT, proving unique end-to-end addresses aren’t critical to the user experience.”
Mobile Traffic and the Rise of IPv4 Extensions
The rapid spread of mobile devices, especially with smartphones, fundamentally altered the internet landscape. When mobile traffic surged after the iPhone launched, ISPs scrambled to expand IPv4 infrastructure instead of implementing IPv6, given immediate needs. “The iPhone and the explosion of mobile traffic created demands no one saw coming,” Huston writes. “IPv6 got pushed aside as providers had to prioritize immediate solutions.”
This rapid shift meant IPv6 fell to a lower priority as ISPs leveraged IPv4 and NAT to keep up with growing mobile demands. As a result, much of today’s internet infrastructure, especially mobile networks, still relies on IPv4.
The Stabilizing IPv4 Market and Happy Eyeballs’ Impact
The COVID-19 pandemic initially raised the price of IPv4 addresses, but costs have since stabilized, now between $30 and $40 per address. Huston interprets this as a sign that IPv4’s scarcity may not be as severe as initially predicted. “If there were really a major shortage, IPv4 prices would be constantly rising,” Huston points out. Alongside price stability, the Happy Eyeballs algorithm—a network technique allowing IPv6-to-IPv4 fallback—has helped keep systems compatible and functional regardless of IP protocol.
“IPv6 may have been the answer for an internet that needed unique addresses for everything. But today, what we’re learning is that’s just not necessary in every case.”
A Pragmatic Approach to IPv6: Coexistence, Not Replacement
Given the slow adoption rate, Huston proposes a balanced model where IPv4 and IPv6 could operate side-by-side, fulfilling distinct roles. “IPv6 was supposed to be the future,” he writes, “but maybe that future includes IPv4 as well.” He envisions a system in which IPv6 can expand as needed without requiring the end of IPv4, especially where IPv4 remains sufficient.
As DNS and CDN routing now play a larger role in managing traffic, IP addresses are less essential to network infrastructure than originally thought. Huston explains, “IPv6 was meant for an internet that needed unique addresses everywhere, but now, what we’re seeing is that it’s not always necessary.”
Related: How to Tell If Your ISP Uses Carrier Grade NAT (CG-NAT)
The term “Carrier Grade NAT”, or CG-NAT, or Large Scale NAT has been thrown around a lot in recent years, but many users are still confused about exactly what it is. In our guide, we show you how you can find out if your ISP uses Carrier Grade NAT (CG-NAT), while explaining exactly what it is and how it can impact your online presence.
Last Updated on November 7, 2024 2:19 pm CET
I think the major issue is that we have sacrificed internet capabilities to maintain it.
By implementing (CG-) NAT, SNI and CDN’s we have essentially abandoned the idea of P2P networking, decentralization and robustness. Now we’re stuck in a situation where even if i WANT to host a website myself, i can’t. Now i need a hosting provider to do that. The entire architecture has changed, and the consequences are entirely not obvious at all.
Take games for example, they need centralized servers to run multplayer features. Basically all data needs to go trough the server because we don’t have P2P connectivity, and this gets expensive. Result is that games either get more expensive, or their servers are shut down eventually, loosing multiplayer access. You could argue that this is a consequence of ISP’s staying on IPv4 instead of doing the right thing and migrating to v6.
And i believe that a lot of people are to blame. Mainly ISP’s of course, but also educational facilities who still don’t teach IPv6 to their IT students, network administrators that don’t enable IPv6 because “nobody needs it”, website maintainers because they don’t think it’s important and platforms like Vercel that have it a low priority. Also don’t forget the browsers, as they are at fault too, since they handle IP mismatches extremely poorly. (They claim such websites don’t exist at all instead of telling users they need to use another network. Seriously, it’s a mess.)
Spot-on.. Getting a dedicated fixed IP address on a standard internet connection now usually requires calling the ISP to make it happen because of CG-NAT. Some even refuse to offer this and you have to work around it with third party dynamic IP DNS services..
Affects also remote desktop.. maybe it’s all a conspiracy organized by TeamViewer, lol
Comment test