Hacker News Reader

Comment by runjake | Hacker News Reader

I was an early adopter implementing and administering IPv6 in a large enterprise, and I still am administering it. It was a shit show for years but in recent years client support has stabilized.

However, I've recently, I've come to my own conclusion that IPv6 is largely doomed and that it will be usurped by IPv4 NATs/CGNATs and >L3 improvements like QUIC, DTLS, etc. Why?

IPv6 is complex and unfamiliar and has a big learning curve. People aren't going to "put up with it". They're going to adapt what they have (skills/tools/etc) and they're doing a good job of it.

I always get asked by people about IPv6 and how they should get started and my present advice is to "hold off, for now". I'm not even sad about that. The world is evolving and when framed properly, it doesn't seem that bad.

View on HN

I find IPv6 to be simpler than IPv4 -- once I stopped trying to reason about things in IPv4 terms. The only real complexity I have encountered is in the various efforts to maintain compatibility with IPv4.

One way or another IPv4 will go away. The limits of scale were hit years ago and the workarounds break applications (NAT is basically a violation of the end-to-end principle and the results are predictable), and merging IPv4 private networks is a nightmare.

I don’t see it. No users care that NAT breaks the end to end principle. That’s only something users care about that want to accept connections, which is <1%.

TCP and UDP have been running over NAT for so long that they work completely fine. People who make the mistake of assuming asymmetric ports or putting IPs in the payload fail so early that they quickly learn.

There is so little actual pressure to implement ipv6 that it’s starting to look pretty doomed.

> I don’t see it. No users care that NAT breaks the end to end principle. That’s only something users care about that want to accept connections, which is <1%.

I'm not sure <1% of users want to video chat or to play games without dedicated servers.

> There is so little actual pressure to implement ipv6 that it’s starting to look pretty doomed.

There is so much pressure to implement IPv6 on mobile that Apple checks if your app can deal with an IPv6 only network before the approve it to show up on the store. If your website supports IPv6 and you have a sizable mobile userbase, you'll see plenty of it.

IPv6 is largely doomed and that it will be usurped by IPv4 NATs/CGNATs...

This argument could have worked ten years ago but now it has been falsified. Cellular carriers have already deployed IPv6 and there's no motivation for them to switch back to IPv4, especially given the high cost of CGN.

IPv6 already succeeded. So far the many corporations already work very hard to perform even internal networks transition. And with new hardware upgrades IPv6 comes for free almost. For example, China recently did a huge leap, where many networks switched at once, along with mobile and web applications. And more is planned in the near future [1].

[1] https://blog.apnic.net/2019/06/06/100-by-2025-china-getting-...

Reply on news.ycombinator.com