What happened to the IP number crisis?

A couple of years ago the internet was going to collapse because we were running out of the old fashioned IPV4 addresses. Well, my router still uses IPV4, though it could use IPV6, and I haven’t heard of any collapsing internets lately.
What happened? Another Y2K?

There was absolutely no “collapse” predicted.

Note that many IP addresses are effectively shared by many devices. E.g., all the devices in my home (and there are a lot of them) share one IP address provided by my ISP. My router has NAT and other tricks to handle all the routing.

A large company can have thousands of computers behind their routers sharing a small number of IP addresses.

The only real issue is what if your ISP runs out and decides to have several customers share an IP address this way. This can lead to issues in terms of functionality, performance and mis-identification of who actually downloaded something improperly. Major ISPs can just buy some blocks from an organization that has a lot of extra to spare.

One college I worked at had a really large block of addresses under its control for a while. Just an amazing number. They foolishly gave up that block for a small subblock in exchange for … nothing at all. The people in charge had no idea of the value.

IPv6 adoption is like fighting ignorance. It’s taking longer than we thought.

The original conception for TCP/IP is that every internet connected device would have a unique IP address that is valid to the world at large. IPv4 had more than enough addresses for the number of systems that were online in the 1980s, but from the 1990s on the number of devices has increased enormously. If we were still operating on the original paradigm, we’d already have run out of IP addresses. People developed technologies to use fewer real IP addresses though, like Network Address Translation that lets you use one address on your router and then give a dozens to thousands of local devices behind it ‘fake’ (non-public) IP addresses, and have them talk to the outside world through the router using its IP address. The workarounds work fine for now, but in the long run there’s still a need for more addresses.

Not sure what you mean by ‘another Y2K’ - There wasn’t a Y2K disaster because people remediated systems that had Y2K problems before the date rollover hit. There was a major effort to rewrite old custom applications and patch commercial applications, and if that effort hadn’t been made then a lot of computers would have stopped working properly when the year rolled to 2000. It also wasn’t just really old mainframe apps, I remember that modern OSs written and sold in the 90s had trouble and vendors had new updrated patches right on into 1999. What we’ve done with IP addresses is sort of like coming up with workarounds that let computers work with dates in the early 2000s, but won’t work past 2025.

Will I need a new router if the new protocol is adopted?

I wouldn’t call it a “new” protocol. IPV6 has been around since the late 90s and was first supported in Windows Vista (for the minor nit pickers, it was technically first introduced in XP in a service pack but was never fully implemented and doesn’t work properly).

Just about anything made in the last 10 to 15 years will support IPV6. For most folks it’s not an issue.

How old is your router?

Does it count as a major or a minor nitpick to point out that neither Vista nor XP was the first to support IPv6? Commercial Unix systems had fully functional and supported implementations as far back as 1997. I’m pretty sure Linux had it even earlier, though maybe in an incomplete or experimental state.

XP came out in 2001 with Vista later than that, and I said IPV6 had been around since the late 90s, so obviously XP and Vista weren’t the first to support IPV6. :slight_smile:

XP/Vista implementation is a better metric for when it started to get accepted on a much larger scale. While Microsoft’s marketing team likes to take credit for just about everything, most internet stuff came from Unix first, especially back in those days.

Linux had it roughly around the same time as other versions of Unix. IPV6 didn’t become a draft standard until 1998 (I think) so anything Linux had before then would have been experimental. A lot of Unix systems had it before then as they were developing the standard.

ETA: I should have said “was first supported ON THE WINDOWS PLATFORM with Vista”. That was what I actually meant. Poor wording on my part. Sorry.

Actually, there was limited support for IPv6 in Windows 2000. It came as a technology preview pack in March 2000. Windows NT got the basics of IPv6 in August 1998. I don’t recall using either.

The biggest adoption of IPv6 currently is cell phones. Nobody really pays attention to what IP address your handset is when its on LTE, so it’s transparent most of the time. I have seen problems reported when there’s some app (often a VPN) that’s v4 only, which means the provider has to send you through some NAT-like 6 to 4 gateway that ends up not playing well if the app isn’t designed to expect it.

A lot of the “problem” was delays due to a chicken and egg thing where destination sites didn’t want to spend the money to adopt IPv6 if ISPs weren’t going to be handing them out, and ISPs didn’t want to spend the money to put in IPv6 equipment if there was nowhere to go. The explosion of smart phones rendered it all moot, the wireless ISPs had to go with v6.

As long as IPv6 is being discussed, who was the “genius” who decided to use colon characters as the address separator? Colon characters are often used in URLs as a port separator, such as:


If mywebsite.com has an IPv6 address of 2001:db8:85a3:0:0:8a2e:370:7334, then the url could be:


The use of colons in the IPv6 address and as a port separator means the URL is harder to parse and harder for a user to see the components to spot an error. For example, did anyone notice that the IPv6 URL above is missing the ‘370’ group? Probably not, since it’s just a jumble of numbers and colons. The IPv4 URL might look like this:

The use of dots in the address makes it much easier to spot what is the IP address and what is the port.

Just need to comment, Y2K came and went without major issues because companies and government organizations took the actual issue seriously and addressed the problem.

The problems didn’t happen as the problems were largely addressed. It actually could have been fairly disruptive. I spent 2 years in 2 different companies making sure we wouldn’t have any major issues. This was done at most companies. There were PC checks, systems checks, telephone system checks, supply chain compliance checks for most larger companies and plenty of oddball systems with integrated processors were tested and corrected where needed. Everything from elevators to dialysis machines were checked and updated or replaced.

I don’t know who suggested the colon specifically, but the intent was to make it easier to differentiate between an IPV4 and an IPV6 address. They also had the intent of embedding an IPV4 address inside an IPV6 address, like this:


Actually, after typing the above I went googling, and I did find out who did it and why. It was Steve Derring in 1992 when he proposed SIP (simple internet protocol) which eventually evolved into SIPP and IPV6. This was the reason given:

Of course running up against the “end” of IPV4 wasn’t going to collapse the internet. Exaggeration for effect. There was a lot of hype about how unless everyone updated to V6 soon the growth of the internet would stall. And of course if services started assuming we have V6 then the V4 apps might not work well. And my question was why these problems haven’t occurred yet.

As for the issue regarding Y2K-I think there was quite a bit of hype involved. Of course a lot of money and effort was spent updating software. But I will point out that the total spent by Italy was something like 1/10th that spent by UK (I am going by memory here but the point about Italy not spending much is the important one), and both countries had the same number of problems-very few. And I believe the night the calendar switched, the US Government reported only 1 system that had a problem. Quite an accomplishment assuming there was a big problem to solve. Forgive me if I continue to believe that in hindsight the problem wasn’t nearly as serious as the warnings indicated.

Actually it dates back to my first DSL line in 2000. It works fine and if it ain’t broke, why fix it? But I will be moving next summer and maybe that will be a good excuse.

If the protocol had been developed as a straightforward extension of IPv4 (as a number of the rejected proposals were), instead of throwing in every unproven idea that was floated, we’d have been there a decade ago. Instead, we got an abomination of a protocol with more misfeatures than features, all to solve a simple problem.

Not true for a number of code-space-limited embedded applications. Neither my air conditioners nor my water heater (for example) support IPv6.

And even if the hardware supports it, your ISP may not. Think that’s limited to smaller ISPs nobody ever heard of? Two words: Verizon FiOS

The two major smartphone ecosystems (iOS and Android) come with development environments that include IPv6 and make it pretty hard to construct an app that doesn’t support IPv6, unless you go out of your way to specifically do that.

It wasn’t just an issue of money for ISPs. Equipment (routers and switches) claimed to support IPv6, but with undisclosed limitations like “punt to processor” instead of being done at the ASIC level, vastly reduced table sizes compared to IPv4 or stealing table entries from IPv4, and so on. And this was major brands like you-know-who, not just the el-cheapo ones. In fact, support was often better on the el-cheapo ones. Certainly the IPv6 ASIC support in LVL7-based devices was better than in Cisco-based devices for a long time.

Most users shouldn’t have to deal with numeric addresses. And the colon was already “taken” by URLs so something else should have been chosen. Just another example of the design group working in the theoretically-impossible “vacuum that is also a perfect echo chamber”. Don’t just listen to me, listen to Randy Bush talking about IPv6 here.

I’ve been an ISP forever. I remember when you could fit the full IPv4 routing table in a Cisco 2500. I have a /24 at home, and right how there are 149 unique IPv4 addresses active in that /24. You can believe me on this, or you can believe the people who say IPv6 is “just around the corner” and have been saying that for two decades now.

Com’on, it was also a major nothing over nothing.

I spent two years telling people “Yes, if it’s not Y2K compliant the date will wrong, why don’t you wait and see?”. And I was also involved with an international company that dropped the ball on a major accounting issue while IT management spend the 2 years ensuring that they didn’t get fired because of a Y2K compliance issue.

Anyway, the idea that there was going to be an IP number crisis turned out to be wrong. And yes, it was sold as a crisis, like the Y2K crisis. But it was always going to be a crisis sometime in the future, and, more or less, for somebody else.

Which turned out to be true. We’ve gone past several critical points in the routing infrastructure, and IP routing has held up, because the central players upgraded their infrastructure. And the routing infrastructure continues to work, partly because the central players have upgraded to IPV6 for some of the critical infrastructure. And we continue to gradually run out of IPV4 endpoints, which continues to be mostly somebody else’s problem, sometime in the future.

For example, I’m a somebody else who may have a problem sometime in the future. My internet connection technology is going to be changed to IPV6 in the next 2 years. Because the upstream routing problems are forcing the provider to IPV6.

When that happens, my current service provider is not currently able to provide an IPV4 static IP address on that channel. So, when that happens, (1) we may change providers, or (2) they may be able to provide the service, or (3) we may decide we don’t need static IP anymore, or (4), we may decide we don’t need IPV4 anymore. Lots of options. Not a crisis for us. But critical business and technical decisions for our provider.

This. A few months ago my home ISP switched us from IPv4 to IPv6 without warning. Suddenly I was unable to connect to my home computers from work and from a lot of public WiFi hotspots when travelling. (And I work in the computer science department of one of the top technology universities in Europe. Despite the fact that we’re known for state-of-the-art research in computer networking, we still haven’t made the transition to IPv6.) I had to call up my ISP and get them to switch us back to IPv4.

My ISP supplies me with a static IPv4 to my video server. Although the ISP does a lot of sharing with IP numbers for the general public – I know hundreds of domains share the same IP – I assume this static IP is not shared. Is that true? If so, it seems like I am squatting on a highly valuable resource for a minor task if you consider The Whole Wide World, and the ISP is only charging $16/month for it (the IP, not the Internet access).

There are roughly 4 billion v4 addresses. $16 a month is a rip-off. That would value the IPv4 address space at 3/4 of a trillion dollars a year. I think the going rate here is $5 a month, and that is as much a token admin fee as much else. It doesn’t really reflect a market value for an address.

One of the things that has limited problems with v4 addresses is that those that have migrated to v6 will often give up their now obsolete v4 address allocations. Friend of mine owns a /24 space. It is essentially worthless.

2nd-tier Aus ISP “Aussie Broadband” is reporting IPV4 numbers each cost around $20 in bulk. If they can continue to give every active connection an endpoint, they only have to use simple IPV4 routing, but it’s gonna cost them around $20 for each customer they add.

Or they could go for NAT or IPV6 NAT, so that they could share a few IPV4 addr among their customers. They’ve done the costing, and they are accelerating their IPV6 roll out and putting in IPV4 NAT in the interm.