Why Doesn’t the Internet Migrate Entirely to IPv6?
Last week at the ARIN 49 Public Policy and Members Meeting we brought together six Internet experts during a special panel to discuss this question: “Why doesn’t the Internet migrate entirely to IPv6?” And wow, did the panel ever deliver an interesting and thought-provoking discussion. Here’s a summary of the compelling conversation and details on where you can watch the full panel for yourself.
Hosted on day two of ARIN 49, the IPv6 panel was moderated by ARIN President and CEO John Curran. He posed four probing questions on the topic that were each answered in turn by these panelists:
- Vinton G. Cerf - Chief Internet Evangelist, Google
- Byron Holland - CEO, CIRA
- Lee Howard - SVP, IPv4.Global by Hilco Streambank
- Geoff Huston - Chief Scientist, APNIC
- Jared Mauch - Sr Network Architect, Akamai Technologies
- Brent Mc Intosh - CTO, MCNET-SOLUTIONS
What does “the Internet Completely Migrating to IPv6?” mean to you, and do you think it will ever happen?
John began the dialogue by getting right to the heart of the issue. Panelists seemed to agree that the migration will happen at some point, but there was little agreement on when.
Vint Cerf said, “Yes, I think it will happen. I think IPv4 will still be around, but IPv6 capability is going to be even more important.”
Geoff Huston concluded, “We’re now doing tricks with names that we never thought possible. Tens of thousands of web hosts sit on one IP address, so all of a sudden, addresses don’t matter. And by the time we start looking, IPv6 will have appeared and that’s the definition of complete migration to me.”
Brent Mc Intosh indicated he thought it will happen, but he couldn’t say when: “Today we’re seeing networks like Akamai, Google, Facebook fully migrated to IPv6, but the challenge that we see is smaller businesses and enterprises and smaller ISPs [Internet Service Providers] are still lagging. … To sum it up for me, full migration is when we have every network running IPv6 natively.”
Byron Holland compared the situation to electrification of the automotive fleet: “Even though city-based electric cars have been around for 30 years, we’re just starting to see that market take shape in a meaningful way, and, even though electric is better, it’s probably going to be a decade more before the fleet is electrified.” He predicted, “We need to be ready for a much longer transition than perhaps this community would ideally like.”
Lee Howard suggested that for the Internet to migrate to IPv6 means that you don’t need IPv4 to usefully connect to the superset of networks that connect to other networks in order to interoperate. He said, “What I’m mostly concerned about is making sure that the things we need to be able to connect to each other can use a common protocol. Probably the long-term best protocol would be IPv6, and so when will it happen? It’s increasingly happening, but I really don’t know. It’s turning out to be a longer tail of migration than I hoped 10 years ago.”
Jared Mauch came right out with a prediction on a timeline: “If the end user can’t tell the difference of what address family, what IP protocol is being used on the other device, that’s when we have truly completely migrated. … I think when you talk about product lifecycles, probably about 25 years.”
What are the hurdles in the way of complete migration to IPv6, and how can we solve them?
John next asked the panel to consider what challenges exist or are foreseen when it comes to complete migration. Vint mentioned most edge equipment has IPv4 and IPv6 capability available, IPv6 just hasn’t been turned on. “One of the big hurdles is getting the ISPs to offer IPv6. For a long time that wasn’t happening. But now ISPs are starting to offer more IPv6 connectivity.”
Geoff pointed to two hurdles: diversity in markets and the price of IPv4. Mobile and consumer markets are embracing IPv6 quickly, but enterprise markets are not moving because there’s no need. As Geoff put it, “It’s a comfortable nondecision to wait, because the perspective is IPv6 doesn’t do it better, it just mitigates future risk.” He went on to say, “The second issue is the price of IPv4. For almost a decade since exhaustion, the transfer markets were pegging the price of IPv4 address really quite low. … [Now, however,] there is an increasing view that this resource is under competitive strain in the marketplace. And quite frankly that strain is the other huge factor that is driving transition.”
Brent focused on ISPs as one of the major hurdles he sees, suggesting that a critical aspect of getting IPv6 on stream is reflecting on the question “What value does it bring [for businesses]?” What is the cost benefit analysis? Challenges for ISPs are carrier-grade NAT [Network Address Translation], IPv6 assessment, replacing hardware, and security. He said, “You have the education, the value proposition, security concerns, and general competitive nature of businesses.”
When asked if Domain Name System (DNS) is a hurdle, Byron replied, “I think it’s going to be the savior, actually.” Geoff nodded and clapped in agreement at this. Byron also thought lack of ISP customer demand and revenue flow associated with IPv6 is a big stumbling block. He looked at the flipside, the organic IPv4 market for addresses that sprang up, and concluded that “until that price escalates to the point that people aren’t willing to pay, there’s not a lot of pressure to migrate … and until that starts to change, I think, the laws of supply and demand will be in effect.”
Lee disagreed on ISPs and said he thinks their linear growth demonstrates fairly good progress. The hurdles he identified were threefold, beginning with enterprise operators who have enough addresses. He followed up with the other two: “I’m not seeing quite the rate of growth in content deployments as I’ve been seeing in eyeball networks, whether it’s mobile or ISPs. But mostly I’m worried about consumer electronics. It’s the devices that nobody is buying based on whether or not it supports IPv6. Why would they? So the real hurdle to me is the complexity of the externalities.” Lee continued, “I don’t think that we’re ready to try and apply pressure to those companies who aren’t going to see a direct revenue or cost avoidance benefit from deploying or enabling IPv6.” Circling back to the question of “when,” he defined it as “when the last current smart TV is out of the last household. … Is it 25 years? Maybe.”
Jared pointed out that he thinks there’s an insurmountable list of hurdles that we need to clear, one of the biggest ones being what he called the “building code” issue. “It’s not just about the technology and the systems that we’ve deployed today. … You have to change the entire mindset around IPv6. You have to change people and how they think, how their software systems work, and how we go about reducing the friction of deploying the technology.”
Does it really matter if the Internet makes it completely to IPv6 and should that even be a goal?
“Let’s go contrarian. … Do we really have to do IPv6?” John asked, pivoting the conversation. Most panelists seemed to agree that there is a need for the Internet to move to IPv6.
Vint responded, “I think it’s still a necessary goal to get there. It would be simplifying the world, because having to run both stacks all the time just seems like an additional burden, to say nothing about the security problems.”
Geoff thinks DNS is the Internet’s future. “Does it matter if we do IPv6 or not? As far as I can see, no,” he said. “It’s just a cost being pushed down the stack to the commodity folk because the real money is in apps and names. [It’s] not in shoveling bits around in addresses. I don’t think we’re going to care anymore in a few years.”
Brent said he “absolutely thinks it’s got to to be a goal,” because he thinks having true end-to-end connectivity is how the Internet is meant to operate.
Byron explained, “In spite of the fact that we’re seemingly in a very, very long transition period, given the number of devices that are coming on, it seems to me that we must have it as a goal.” Noting that there are concerns about performance and effectiveness of IPv6 failing over to IPv4 at unacceptable rates, he said he doesn’t see an extended transition time as a bad thing.
Turning the floor over to Lee, John expanded on his original question: “If we do [need to do IPv6], is an efficient market getting in the way of that? Lee began by countering the question of whether IPv6 should be a goal with his own: “For Whom?” He said, “When I was working at an ISP it was absolutely a goal for us, because we knew we were going to run out of IPv4 addresses. … In that case, an efficient market was absolutely a countervailing force that drove us towards IPv6. But there are nonmarket participants, like enterprises and consumer electronics folks, who don’t feel the pinch of either market prices or of running out of addresses. So there’s less incentive for them.” He continued, “Although, IPv6 has lower latency than IPv4. I’ve talked to some companies that have said, ‘That’s really exciting. We were not concerned about buying addresses, but if we can improve our performance by 15%, that’s fantastic.’”
Jared summed up the discussion with his comment, “We absolutely have to get to IPv6. It is really the future of where we’re going.”
As a Regional Internet Registry (RIR), what should ARIN do about IPv6 migration?
Noting that while ARIN is involved in addresses and its philosophy historically has been “do what you want to do,” John opened up the question of our role in IPv6 migration.
Vint said, “I think ARIN is doing the right thing, which is enabling parties who need address space of any kind to somehow get access to it.”
Geoff thinks that “the five RIRs should be able to accurately, at all times, where addresses are— who’s got them, all the time. If you do that, that’s 95% of the job.”
Brent said ARIN should just continue doing what it’s doing: “You are supporting and engaging your stakeholders with your IPv6 deployment stories and publications and supporting/publicizing what other companies have been doing with IPv6 migration. It’s a big deal. Recently we have learned a lot from it.” Emphasizing that making it easy to apply for and get address space is important, he continued, “Secondly, I think you have a really, really great supporting team in ARIN.”
Byron said, “First and foremost … run a highly effective, efficient, and bulletproof registry, which includes who has what, where, and when and the ability to register, transfer, etc. So to me, that’s job number one. I think that there’s also an opportunity potentially for a little more activist role in terms of education and awareness.”
Lee appreciates that ARIN requires IPv6 for its services and noted that AFRINIC has done a couple of IPv6 deployathons. He also said, “There’s still a surprising amount of open-source software that does not support IPv6. … Maybe ARIN could either sponsor or provide some testbeds or developers.” Jared reiterated, “ARIN should be focused on its core function, which is maintaining the registry of the numbers that have been subdelegated to it from IANA [Internet Assigned Numbers Authority].” Beyond allocating out of those pools, not much.
An IPv6 packet walks into a bar. No one wants to talk to it.
Jokes weren’t the only thing shared in the comments on the floor and in the online chat during the panel session. The conversation amongst ARIN 49 attendees touched on quite a few serious points as well. People shared comments in the vein of:
- The old gear problem is a lot bigger than people think it is. The cost of replacing equipment is expensive.
- Pricing for maintaining IPv4 connectivity is not felt by those writing the software.
- Training people is another concern.
- Workarounds for maintaining IPv4 like NAT/Carrier Grade NAT is invisible to most people, but if the workarounds had more pain points, more people would notice.
- An example was shared about requesting IPv6 capabilities for corporate connectivity: Even if it is not a priority for a specific project, that can start to place some pressure for ISPs to consider it as an option.
- Unfortunately, IPv6 is also the first thing people turn off if they have any network issue. It’s troubleshooting step number one, and then it never gets turned back on.
- As an industry, we should make sure we don’t treat IPv6 as an add-on, but rather as a building code requirement for the Internet today.
You get IPv6 Address, You get IPv6 Address, You get an IPv6 Address
After asking his set of initial questions, John took questions from the audience both in person in Nashville, Tennessee, and from online via Zoom. Some of the conversation that resulted from opening the floor included asking for the moderator’s vision for the future of IPv4. John pointed out that ARIN’s job number one is to keep an accurate registry. He said, “This is your ARIN, and you’re the community. It’s actually you who tell me that vision.” In response, Jared suggested that when anyone comes to transfer IPv4 space, ARIN could say, “By the way, here’s your IPv6 on the side, as well, for free.”
Another conversation point revolved around the 25-year transition prediction that didn’t seem to consider the need to keep refreshing, especially in terms of Internet of Things (IoT) devices that people bring into their homes. Panelists also offered their opinions on whether they think we’ll continue to see the transfer market sustained until the majority of IPv4 space is in use and whether they think the market affects IPv6 adoption. John Curran summarized the conversation, “We’ve unleashed the market, and we have to realize now we’re the tail on the dog.” Then the final two questions to panelists were: Do you think IPv4 is only a technical debt issue and, if so, how do we get companies to address the problem?; and lastly, do you think the US government requirements for IPv6 will have an impact at this time?
To hear the responses to those final questions and to watch the full panel for yourself, check out our webcast archive below. Then let us know your thoughts about the Internet’s migration to IPv6. You can find us on social media at @TeamARIN.
Recent blogs categorized under: IPv6
GET THE LATEST!
Sign up to receive the latest news about ARIN and the most pressing issues facing the Internet community.SIGN ME UP →
Blog CategoriesGrant Program • IPv4 • Public Policy • Elections • ARIN Bits • Tips • Fellowship Program • Updates • RPKI • Outreach • Training • IPv6 • Internet Governance • IRR • Data Accuracy • Customer Feedback • Caribbean