Home / Blogs

The Irrationality of Deploying IPv6

For the past few decades, there’s been a relatively straightforward narrative on the economics behind the IPv6 transition that goes something like this: sooner or later, IPv4 scarcity will drive costs up until they exceed those of deploying IPv6. A competitive market will then make the rational choice and transition to a more efficient mode of production and deploy IPv6. This is textbook economics, and—with the disclaimer that I’m not a trained economist—it appears to be incorrect.

Before we look at what’s wrong with this, let’s examine some of the baseline facts, starting with what underpins it all. IPv4 allows for a maximum of 4.2 billion unique addresses which, after some technical constraints and reserved blocks of addresses, leaves around 3.7 billion addresses which can be used on a global and unfragmented Internet. This means that you can connect 3.7 billion ‘things’ to the Internet. But no matter how you count, be it humans, connections, or indeed actual things, we can be sure by now that we have exceeded this number. And as the Internet keeps growing and we keep connecting more people and things, this problem only gets bigger.

The limitations of IPv4 formed the basis for developing and standardising IPv6. This work was completed in the second half of the nineties, before the dotcom boom when the Internet really took off. Using an address field of 128 bits, IPv6 expands the number of unique addresses towards 340 undecillion, which should be enough for the foreseeable future. With the protocol written and standardised, the rest should be up to the market. Or at least, that was the assumption based on the notion that, within the bottom-up model of the Internet, the market gets to pick and choose which standards to use and implement.

Knowing that there was a very real and hard limit in IPv4, we’d all assumed the market would soon recognise IPv6 as the more efficient protocol, allowing the networks and services to continue growing beyond the barriers of IPv4. As a former engineer, I know assumptions are dangerous and the case at hand appears to be a textbook example. It has been two decades since IPv6 was invented and the Internet is well and truly exceeding the number of users and devices that the IPv4 address space allows for. The consensus view at the moment is that only about one third of Internet users can use IPv6. Can use, because there are two sides to every connection, and you can only use IPv6 if both sides support it. If that is not the case, the fallback is to use IPv4.

This is where the problems start to appear, because as much as I can have a fully functional IPv6 setup, I need IPv4 as a ‘plan B’ in case the other side of my connection doesn’t. With the global and regional pools of available IPv4 addresses depleted, globally-unique IPv4 addresses have become a scarce resource. Unfortunately, this where the initial thinking got it right: a scarce and limited resource, with demand sustained or even growing, will go up in value, making it more expensive. Indeed, the value of IPv4 addresses has risen sharply over the last few years.

Faced with a dwindling supply of unassigned IPv4 addresses, people quickly figured out that there is always a certain degree of overhead in IP address allocations. Due to being binary, address blocks have a fixed length, so even if you have 129 devices on your network, the minimum size address block is 256, and the assignment process and policies also took some level of growth into account to prevent the address space from becoming too fragmented. The surplus within these assignments quickly led to the development of a secondary market in allocated but unused IPv4 addresses.

As demand grows and supplies go down, the price is going to increase. It was a bit of an outlier, but last month a small block of 256 addresses was sold at auction for just over 59 USD per address. An exceptional price maybe, but there is a very clear trend of prices going up. Although the policies at the Regional Internet Registries contain protections against speculation—once a transaction is processed, there is a “lock in” period where the same address block can’t be transferred again—there are signs that arbitration is taking place and prices are starting to converge towards a single global level.

The price of IPv4 is going up, but will it lead to large-scale adoption of IPv6? Is the market rational enough to realise that there is a more efficient means of connecting things and people to the Internet? We see lots of networks and services deploying IPv6, but it is not immediately clear that there is a massive competitive advantage for doing so. In fact, the case can still be made that there is some disadvantage to enabling IPv6, knowing that you must also maintain backward compatibility and bear the cost of maintaining connectivity to IPv4 or even expanding on your IPv4 address space. To some extent, running two protocols doubles your effort and incurs some extra costs.

If we were faced with a truly rational market, and the model I described in the first paragraph holds true, then a lack of IPv6 deployment indicates that we have not yet reached the tipping point. For the average market participant, sticking with IPv4 might still be cheaper than deploying IPv6. This also doesn’t sound right though, because running a single protocol should be cheaper and much easier to do.

The basic product ‘Internet access’ has become a commodity, and at the national level, many markets have already reached saturation point or are close. The main factor that sets competitors apart is price, and with the associated drop in profit margin, you’d expect a more cost-effective solution (IPv6) to gain traction very quickly—but it isn’t.

So, if it isn’t price, what would it take for the world to deploy IPv6 to a level where IPv4 support is no longer necessary? Because there likely will be a tipping point where a cost/benefit analysis shows that it’s more efficient to switch IPv4 off than to continue maintaining it. For such a scenario to happen, there has to be some coordination between the different actors. Enabling IPv6 is a decision you can make yourself but using it to transport data across the Internet requires the other side and any intermediaries, such as wholesale connectivity providers, to make the same decision.

This touches on what I think is one of the fundamental properties of what makes the Internet what it is: no matter how fierce the competition, you need to work together. Without a top-down structure, as we saw, for instance, in traditional telephony, coordination and cooperation between networks is what keeps it all together. The market picks its protocols and standards, but to truly benefit from the network effect where value increases with the number of participants, there needs to be some consensus on which protocols are the winning ones.

This is what sets the Internet apart from other networks and what makes the case for IPv6 a very challenging one. Because, at the technical level, there is broad consensus that IPv6 is the winner, and that consensus has been there ever since the protocol was standardised (which was itself a consensus decision).

Finding the same consensus in the boardrooms of today is a different matter. This is because while it’s easy to position IPv6 as the solution for long-term growth and a more efficient means to run the network, making the financial numbers work is much harder. There is a gain for sure, but it can take years for this to translate into profit, and there are a lot of uncertainties and externalities along the way. Especially when there is a risk-averse management or a focus on the next quarter or the end of the financial year, it can be tough to make a business case for deploying IPv6.

What is needed most for IPv6 deployment to become the de facto way to move data across our networks is vision. We need executives to share the vision that engineers had long ago, of a truly global network with room to grow. A network that not only allows you to connect every person and thing on this planet, but one that supports innovation and has the flexibility to adapt to new challenges and stand the test of time.

The IPv4 protocol has gotten us a long way, but with its limited number of addresses, it has run its course. Connecting more people and new things is getting harder, and we need IPv6 if we want to continue to expand and grow the Internet and its capabilities.

Is there another way? Perhaps there could be some intervention, for instance, by legislators forcing network and service providers to deploy a particular protocol. We certainly can’t rule this out. Many governments and even the European Union are considering their options here. For instance, the European Commission’s cybersecurity strategy mentions that it won’t exclude regulatory measures ‘to steer the market’ if it thinks insufficient progress is made.

While this can’t be ruled out, and we can discuss whether such measures would be effective, we can’t ignore that such an intervention would alter the fundamental principles of the Internet. Network operators are free to choose whatever protocols and technologies they think are best or that bring added value. This fundamental property of ‘choice’ is what got us where we are today. It is likely a big part of why the Internet has become such a success and why it has mostly overtaken all other forms of telecommunication and data networks.

Politicians talk about maintaining the Internet as a global unfragmented network that supports and lets us enjoy our freedoms, like they did in the recent G7 declaration. IPv6 provides a path to make that happen, but we all need to do our fair share. The Internet was intended to be an open, collaborative structure of networks. If we all invest just a little, for instance, by enabling IPv6, we can maintain this vision and keep the network a free, global and unfragmented space. Because in the end, the true value of the Internet cannot be expressed in monetary or economic terms alone; it is so much more than that.

By Marco Hogewoning, Manager of Public Policy and Internet Governance at the RIPE NCC

Filed Under

Comments

I've wondered about the real deployment numbers, Todd Knarr  –  Aug 3, 2021 8:45 PM

I’ve wondered about the real deployment numbers, because I’ve noticed that Windows systems (we can safely say that’s the majority of consumer users) prefer IPv4 over IPv6 when both are available (and they often are, Windows sets up IPv6 automatically and uses it for internal and LAN purposes even when the primary external connection is IPv4). That results in mostly IPv4 connections to the rest of the Internet, IPv6 only gets used when one end or the other simply doesn’t support IPv4 at all (very rare on the consumer side except for cel networks, pretty rare on the server/provider side).

My thinking is to watch the big cloud providers and CDNs everybody uses (AWS, Azure, CloudFlare, etc.). When they start offering external connectivity with IPv4 as an extra-cost option, not a standard part of the configuration, that’s when we’ll be at the tipping point.

Windows and every other operating system prefers Kenyon Ralph  –  Aug 3, 2021 9:07 PM

Windows and every other operating system prefers IPv6 over IPv4 when both are available. Your Windows system preferring IPv4 must not have global unicast IPv6 addresses available. If it only has unique local addresses or link-local addresses, IPv4 would be preferred, but for regular global addressing, IPv6 is preferred (notwithstanding Happy Eyeballs and other application-level nuance).

(6to4 and other transition technology IPv6 addressing Kenyon Ralph  –  Aug 3, 2021 9:09 PM

(6to4 and other transition technology IPv6 addressing might also be less preferred than IPv4, if that's all you have.)

We've got native IPv6 available and active. Todd Knarr  –  Aug 4, 2021 3:14 AM

We've got native IPv6 available and active. One thing I haven't checked though is whether routers these days are enabling IPv6 by default. I'd thought they were, but it's been so long since I've used a router that wasn't reflashed to DD-WRT...

Deployent vs. Traffic Abraham Chen  –  Aug 14, 2021 7:55 AM

Dear Todd: 1) "real deployment numbers": This is closely related to "IPv6 Capable" or "equipment readiness" statistics that the best place for the latest data is at: https://stats.labs.apnic.net/ipv6 2) However, the real measurement of the IPv6 deployment status is its traffic volume. That is, the percentage of the entire Internet activities that is carried by IPv6. For this, looking at a particular sector of business or a few major companies may not tell the true story. There was a discussion on this forum that presented the in-depth up-to-date (back then) information that you can review: https://circleid.com/posts/20190529_digging_into_ipv6_traffic_to_google_is_28_percent_deployment_limit/ Regards, Abe (2021-08-14 10:54 EST)

USG IPv6 mandate Einar Bohlin  –  Aug 4, 2021 6:09 PM

Great article Marco. Hopefully USG IPv6 mandate will continue to spur action, and not only within the federal government. Per OMB memorandum M-21-07 the first milestone is in 2023 and the third milestone will require that, “At least 80% of IP-enabled assets on Federal networks are operating in IPv6-only environments by the end of FY 2025.” That’s said to be on track. However, at this time cyber security is clearly first on the list of priorities, and that will have an impact.

Sort Out the Technical Aspects Before Considering Economics Abraham Chen  –  Aug 5, 2021 8:26 AM

Dear Marco:
0)  I agree with your starting point, but not quite in sync with the line of your thoughts about seeking governmental intervention, because there are at least a couple technical issues that IPv6 needs to sort out first for the Internet to be a solid world-wide communication infrastructure deserving such attention:
1)  IPv6 started out to be not backward compatible with IPv4 that any large system planner knew as the very first requirement. If IPv6 design had this in mind, the transition from IPv4 could have been so stealth that it was a behind-the-scene operation from outsiders’ perspective, just like how traditional PSTN has done in improving telephone services thru the decades if not a full century. Lacking the backward compatibility, IPv6 rollout is handicapped by the fact that the majority of the environment is still IPv4. The latter is a fact that everyone relies upon everyday. There is no compelling reason for anyone to be the first to forgo such.
2)  By extending the well-tested RG-NAT (Residential Gateway - NAT) technology into CG-NAT (Carrier Grade - NAT), the current Server - Client model became the primary Internet operation configuration, relieving the IPv4 address shortage issue. There is no need for IPv6 addresses for now.
3)  On the other hand, IPv6 has not been promoted with its strong suit of enabling end-to-end communication between individual IoTs. If this were made popular for the mass public to enjoy their individualized imaginations, there would be strong demand for IPv6 from end-users, countering the current Server-Client model.
4)  From these considerations, it is no surprise that a recent APNIC blog classified the current situation as “the reason why IPv6 is still seen as an option, rather than an urgent necessity.”

  https://blog.apnic.net/2021/07/21/another-chapter-in-the-decline-of-the-traditional-telcos/

Regards,


Abe (2021-08-05 11:25)

Point one is on point The Famous Brett Watson  –  Aug 8, 2021 12:47 AM

The point about the lack of backwards compatibility (or some other sane migration strategy) is one I don't hear nearly often enough, which makes me think that no lessons have been learned. This isn't a failure of the market: it's a failure of engineers to deliver the product the market really wanted. I chalk it up to hubris: flush with victory at being the de facto standard (beating out CLNP) and a possible case of second-system effect, they seemed to think that they could design the next iteration as though from a clean slate and the world would adopt it simply because it bore the "Internet" imprimatur.

Monopoly Does Not Work Here Abraham Chen  –  Aug 14, 2021 7:12 AM

Dear Brett: You are identifying a very intriguing aspect of the Internet. That is, it promotes diversity and decentralization. In reality, however, much of the current setups and conventions in the Internet are going just the opposite direction. In the case of IPv6, the designers' mentality was monopoly. So far, the real world at large has not bought into it. Regards, Abe (2021-08-14 10:11 EST)

Todd Knarr  –  Aug 13, 2021 11:05 PM

Another point in favor of pushing IPv6: the current Netflix VPN blacklist hitting residential IP addresses in significant numbers. It used to be that when VPN providers abandoned flagged blocks of address space in favor of newly-acquired “clean” blocks, you could count on those abandoned blocks not being reallocated quickly. Services like Netflix didn’t need to update their blacklists to remove now-abandoned VPN blocks on a daily basis, they could safely leave it until consumers complained. Now, though, with IPv4 address space getting tighter and tighter those abandoned blocks are going to get allocated to another company fairly quickly and the lag before reallocation is only going to get shorter and shorter. Companies like Netflix are going to have to either actively monitor address space activity and update their blacklists as blocks are released or face a situation where by the time they notice any consumer complaints about incorrect blacklisting it’s going to be headline news on the tech sites.

Comment Title:

  Notify me of follow-up comments

We encourage you to post comments and engage in discussions that advance this post through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can report it using the link at the end of each comment. Views expressed in the comments do not represent those of CircleID. For more information on our comment policy, see Codes of Conduct.

CircleID Newsletter The Weekly Wrap

More and more professionals are choosing to publish critical posts on CircleID from all corners of the Internet industry. If you find it hard to keep up daily, consider subscribing to our weekly digest. We will provide you a convenient summary report once a week sent directly to your inbox. It's a quick and easy read.

I make a point of reading CircleID. There is no getting around the utility of knowing what thoughtful people are thinking and saying about our industry.

VINTON CERF
Co-designer of the TCP/IP Protocols & the Architecture of the Internet

Related

Topics

New TLDs

Sponsored byRadix

Cybersecurity

Sponsored byVerisign

Brand Protection

Sponsored byCSC

DNS

Sponsored byDNIB.com

Domain Names

Sponsored byVerisign

IPv4 Markets

Sponsored byIPv4.Global

Threat Intelligence

Sponsored byWhoisXML API