You are here: Home > Publications > RIPE Labs > Marco Hogewoning > Can You Make IPv6 Work Commercially?

Can You Make IPv6 Work Commercially?

Marco Hogewoning — 07 Jul 2016
Large scale IPv6 deployments suggest that IPv6 is at least a technical success - the technology works. Now it's time to visit the other important question: does it work commercially? Does IPv6 really come with a positive business case? We are about to find out, if you help us... (for instance by filling in the poll next to the article)

 

Our technical community has spent about two decades making IPv6 work on a technical level. We have developed the protocol, modified and expanded a few others; we set up the registry system and distributed the addresses. In addition, over the last 10 years we have invented pretty much any possible way to encapsulate or translate IPv6, making it easier to integrate with the IPv4-based world we still live in. And we have succeeded: when in Belgium, there is about a one-in-two chance your Internet connection supports IPv6; on a global scale, Google (on a good day) sees one-in-eight customers connecting via IPv6.

Are we done then? After all, we can show that IPv6 works, even on a massive scale with millions of users. We have written all the documentation there is to write, we have educated and trained all of our colleagues and even created awareness outside of our own community about the need to transition the Internet to use IPv6. Meanwhile the IETF has already taken steps to investigate and discuss the consequences of the inevitable "shutdown" of IPv4. All we need to do is sit back, relax and wait for the IPv6 transition to complete, which is just a matter of time.

Or is it?

As part of my job, I spend a lot of time talking to people about making the Internet better, and the need to deploy IPv6 is part of that conversation. I have had such talks with small operators who came seeking some more IPv4 addresses.  We have had similar talks at the highest level of the United Nations, where the question on the table is about getting the rest of the world connected. Lately, with the Internet of Things at the top of the expectation curve , talks have shifted from connecting people back to connecting machines, and again IPv6 is flagged as a basic requirement to make it all work.

“Why haven’t you enabled something on IPv6 today?” I am paraphrasing here, but essentially that is the question we mostly struggle with today. The awareness is certainly there and there is, with a few exceptions, quick agreement that IPv6 is the only way to continue expanding the Internet and protect the values we all cherish so much. Yet in many countries and countless networks there are few if any signs of IPv6 use.

Why are you not deploying IPv6? We can pretty much rule out technology. We can show cases with millions of users active on every mainstream technology out there. If IPv6 works for them, why doesn’t it work for you? The other thing often cited is business risk, for instance an increased level of support calls. Admittedly all new beginnings are hard and while it might be proven technology, you still need a solid plan for your IPv6 roll out, which should include lab tests and a slow start to be able to handle any unforeseen circumstances. But the bottom line is, if one-in-eight visitors to Google use IPv6, it certainly can’t be that bad. If IPv6 were that fragile, I would probably not be writing this article, but more likely be out somewhere explaining what went wrong.

That leaves me with the most commonly cited reason: money. Just as with any other change, there is always some cost involved. Any well-thought-out plan takes time to write, test and implement, and you probably won’t do it for free. Which leaves the question: how much does it cost to deploy IPv6? That is a really tough question to answer, and there is hardly any reliable data available that can help us.

From what is available, it seems unlikely that there is a correlation between a country’s economic performance indicators (such as the GDP) and the level of IPv6 deployment. Certainly we see a higher IPv6 adoption rate in the developed world than we see in the developing nations of the Global South. But there are outliers there, especially in South America. Even within the European Union there is a large difference between adoption levels in countries, and it doesn't line up with the size or strength of economy.

At the carrier level it is near impossible to determine the actual operational and capital expenses that are involved. Such figures in general are considered company confidential and all I can say is that I am not aware of any publicly-listed company detailing the costs in one of their annual or quarterly reports. Which makes me think it can’t be that high. Those operators who have deployed IPv6 at a large scale are still all in business and are still accountable to their shareholders, so I suspect that a positive business case can be made.

Others have tried this in the past, but what we can do is create an estimate of the cost of not deploying IPv4. After all, we have a rough indication of how much IPv4 addresses are traded for in the secondary market (around 10 USD). And we can take an educated guess on the cost of sharing such an address among multiple customers using a NAT (which on a very conservative estimate over a 3-year period is also about 10 USD per customer). Judging by the transaction volume of IPv4 transfers, for a lot of companies these costs are manageable from a business case perspective. Arguments that expanding business on IPv4 is a short-term affair and not very future proof notwithstanding , the fact that these transactions are taking place is a firm signal that there still appears to be a business case for IPv4.

As part of this year’s inter-sessional work for the Internet Governance Forum (whose 2016 event will be held in Mexico this December), a group of volunteers has picked up the daring task of trying to describe the commercial and economic reality that underpins a successful deployment of IPv6. As part of the project to document IPv6 best practices, we are hoping to gather some input on the costs and (hidden) benefits of IPv6 deployment that lead to a positive business case and that will convince the product managers and boardrooms who are now stuck with the challenge of expanding their business using a finite and very much exhausted resource, to deploy IPv6 within their products and services.

Can you help us? Share how you make IPv6 work in a competitive market, share the arguments behind your business case - maybe it was just a matter of your competitor deploying it? Even if you haven’t deployed IPv6, please share your arguments or business case, as this would also help us to gain insight in what is happening here.

More information about the IGF Best Practices Forum will soon be posted at http://www.intgovforum.org/cms/best-practice-forums/bpf-ipv6 from where you can also subscribe to a dedicated mailing list. Of course, as RIPE NCC, we are also happy to collect and take forward any feedback you might have.

4 Comments

Pier Carlo Chiodi says:
14 Jul, 2016 03:14 PM
Just to further enrich the IPv4 address sharing costs topic I want to link here a tweet from the last Swiss IPv6 Council event (probably many of you have already seen it):

    https://twitter.com/perfexcellent/status/743341303346274304

"Swisscom found transport cost of 1Gbps is 8 times more with NAT than native IPv6"

"Is this a business case or not?"
Robert McClean says:
09 Sep, 2016 06:23 PM
When you are an ESP you need to connect with mail exchange servers all over the world in order to deliver emails. We have found that very few systems will accept IPv6 connections (Google are an exception). This leaves us with dual-stacking which takes extra time, and time is money. When it comes to sending the email, the only way to find out if IPv6 is supported is to attempt to connect using IPv6 then fall back to IPv4 if that is unsuccessful. This slows the process down to the extent that more server capacity is required, which costs money. These are the obstacles we face in adopting IPv6.
Marco Hogewoning says:
12 Sep, 2016 10:23 AM
Just to clarify, but with "testing if IPv6 is available", you mean look for an AAAA on the MX, right?
Robert McClean says:
12 Sep, 2016 12:11 PM
Partly. An AAAA record on the MX indicates the protocol is accepted but is not a guarantee that it will allow the email to be delivered. This is what we found, though things may have changed in the last year or two.
Add comment

You can add a comment by filling out the form below. Comments are moderated so they won't appear immediately. If you have a RIPE NCC Access account, we would like you to log in.

IPv6 Return on Investment Poll

Loading poll

When do you expect a return on investment for your IPv6 deployment?

 We're seeing it already

 Within one year

 2 - 5 Years

 In more than 5 years

 Never