Big rise in IPv6-enabled .nl domain names | Hacker News Reader

Big rise in IPv6-enabled .nl domain names

sidn.nl

37 pointsokket14 comments

The real question being when we will be able to switch off IPv4? As long as we don't, the pain associated with the IP shortage will have to continue to preserve backward compatibility. So really IPv6 will not actually solve the problem for a very long time.

There are a few different goals here:

* When can consumer ISPs turn off v4? This can and does happen right now - the ISP just needs to have a carrier-grade NAT gateway to handle connections to IPv4 servers. This is a performance problem if most services are only available on IPv6, but we're already at that point.

* When can service providers and hosting companies shut off IPv4? This is the problem you address, and that's gonna take a looooong time.

* When can ISPs shut off their IPv4 gateways? When all services (within reason) are available on IPv6. This one is of less interest than the other two, though.

I would have thought it would work the other way? Surely ISPs will have difficulty switching off v4 access to consumers, because of all the old v4 devices out there. But at some point soon the vast majority of servers will support v6, a which point some big ISP will say 'we will no longer support v4-only servers after date X, we will NAT to v6 for comsumer devices if you want to talk to our customers you need v6'. Some time after that IPv4 will effectively no longer be a global protocol.

You have the technical answer, but I think the more useful answer is human based.

Content providers will stay dual stack basically indefinitely - abslutely no website operator wants to lose even 1% of their traffic due to a technical limitation, so I do not expect demand for v4-enabled websites to die off any time soon. I would expect 95%+ v6 usage before you even see this being a thing commercially, and even at those numbers only very small marginal operations will go v6 only.

The consumers and consumer ISPs have both a better technical solution, as well as no consumer ISP gives a shit if their customers can't access 2% of the internet. Therefore the change due to both technical reasons and simple game theory must come from the "eyeballs" side of the equation.

The IPv6 shutoff is for clients first because NAT is a solution only for v6 client to v4 server; the way it works is you send packets to an IPv6 address that has an IPv4 address stuffed into 4 convenient bytes, and the NAT gateway installs a mapping for that communication. This doesn't work the other way around, because there's no way to address the IPv6 internet endpoints in an IPv4 packet to set up that initial mapping.

Generally the limitation is more about infrastructure (including private in-home routers) than about endpoint devices; Windows, OSX, iOS, Android, and Linux have all supported IPv6 for years.

I don't believe there is a need to completely abandon IPv4; at least for small internal networks it is probably easier -at the moment- to quickly set it up as IPv4. A properly configured IPv6 network, however, should be equally easy to use. If DDNS is working correctly then there is not even a need to remember IPs. The problem arises from poor support in consumer equipment. At least for most mid-range routers I get the impression that IPv6 support is hacked together as an afterthought rather than be considered from the original implementation of the firmware. So you get something that mostly works but it's not 100% there. It's getting better but we are not there yet.

"We" meaning who?

IPv6 has already simplified my life, when I asked "hm, there aren't enough free IPv4 addresses there to set up <stuff>, do <organisations> all have IPv6?" and got a positive answer back.

If you only cater to a limited set of specific clients who do not move around the internet, you might get to that conclusion. But if you need to be accessible from the general public and/or your clients may connect from anywhere in the world (mobile/public wifi/home/etc), you will likely have to support IPv4 and architecture your solution around it.

Sure. Sometimes software needs to serve a) one or a few organisations and sometimes b) everyone. IPv6 alone is far from sufficient for point b.

I guess everybody also has to actually use IPv6 then. I must admit I'm guilty myself: whenever I set up a VM, I just check the v4 connectivity. When setting up Network stuff, the same holds true.

Sure the one or other device/service doesn't work with it well, but maybe that's a reason to not use that. Linux users also avoid Hardware that only goes well with Windows.

ipv4 will die approximately never, but it might not be useful for most applications. the roles of v6 and v4 will simply swap.

Once v6 reaches the same adoption of v4 then it'd make sense to transition v4 out. I'm guessing addresses will be released and not redistributed again, eventually completing a total recall of v4.

We're still 2-3 decades at a minimum from that milestone though.

> For the last six months, we've additionally been rewarding registrars that have a lot of IPv6-enabled domain names in their portfolios. And the policy is working!

Ok, so there is an incentive...that explains it. Who knows what the "reward" is, but the increase is pretty impressive. Kudos to the Dutch for their increased IPv6 adoption!

> the number of IPv6-enabled .nl domain names has more than doubled in the last six months.

Clickbait headline?

Reply on news.ycombinator.com