Comment by wredcoll
3 days ago
> IPv6 should have really just worked like NAT/IP over IP on IPv4 hardware
This literally exists. There's no need for it to be part of the original specification because it's a completely separate concept.
There's nothing wrong with ipv6 other than the general human desire to not change things that "work" and for people who are currently powerful because they own a bunch of ipv4 to resist changes.
> There's nothing wrong with ipv6 other than [long list of human things]
The unfortunate reality is that technology exists to serve us. If humans don't want to use it, it means that your technology is wrong. End of story. It doesn't matter if you (or any of us) think it's elegant and beautifully designed.
That's like blaming the photovoltaic cell for the lack of solar power in america.
There's no way to design a technical solution to a social/human problem at ipv4 shortage is a human problem.
People have spent literally the last 40 years trying to come up with a solution "better than ipv6" and NOBODY HAS. Does that tell you anything?
Declaring a technology-in-transition dead is truly a take.
It's like declaring SSL/TLS dead in 2012 because some websites were not using it. (Yes, I remember the dark times when even authentication data was transmitted over plain text.)
People still mine, ship and burn coal for energy.
Turns out its hard to get people to change their own habits to help other people.
My first job out of college passed the authentication over SSL/TLS, but the page that *contained* the login form? That was sent in the clear.... Truly dark days.
If your solution is that “if everyone just used IPv6 we wouldn’t have this problem” then it’s not a solution. Standards like USB, Wi-Fi, HDMI, DP, etc. are so successful because they are by out-of-the-box backwards compatible with their predecessors.
>HDMI, DP, etc. are so successful because they are by out-of-the-box backwards compatible with their predecessors
Takes on Hacker News can truly be wild sometimes.
I certainly don't recall being able to use DP on a VGA monitor without some sort of adapter!
You missed the part where DP didn't come with it's own Audio Codec and GPU architecture. Of course, I am exaggerating, but the point is that the problem with IPv6 is not just that it suffers from bad design on merit, it also critically suffers from scope creep. We wanted a larger IP space, instead we got network discovery and globally unique cookies as a bundle deal, with a whole lot of other crap.
2 replies →
> This literally exists. There's no need for it to be part of the original specification because it's a completely separate concept.
By that token, IPv6 never had to exist at all then, but we know that is completely missing the point.
IPv6 is a privacy nightmare without privacy extensions. It is designed with "a stupid desire to pretend we can boil the ocean".
Close to 3 decades and we are barely scraping 50%: https://stats.labs.apnic.net/ipv6/
Running a IPv6-only network that doesn’t interact with IPv4 networks in any way is great and it works.
The actual gap is the mess of mixing/transitioning, and it wasnt planned properly in the last 30years
Yeah, that is the whole point. It was never going to be a clean cut. "Works in my lab" isn't good enough for real world.
The issue isn't that it wasn't planned properly, the issue is that the real world wasn't taken into consideration by people who either worked for the likes of Google who were aching for a globally unique identifier per device or just delusional about how much we can change in real world.
6 replies →
> without privacy extensions
Privacy extensions are enabled by default on every major operating system.
> There's nothing wrong with ipv6
Sweet summer child...
There's EVERYTHING wrong with IPv6. It's a recursively bad protocol. From the basic header design (extension headers, wtf) to the "human management" layer ("why should we deploy IPv6 if it actively _breaks_ stuff?").
As an example, try to design a standards-compliant IPv6 network where you can reliably print from an Android phone if your Internet connection goes down and the router withdraws the prefix. It's trivial with IPv4 and _impossible_ with IPv6.
> if your Internet connection goes down and the router withdraws the prefix
My router advertises both the public prefix from my ISP and a ULA prefix for internal LAN connections (essentially the IPv6 equivalent to RFC-1918). This gives me stable local addresses that do not change and work without public internet access.
https://en.wikipedia.org/wiki/Unique_local_address
> My router advertises both the public prefix from my ISP and a ULA prefix for internal LAN connections
You'd think so, but you'd be wrong. If you have a GUA and a ULA, and the printer has both a GUA and a ULA, then your host is going to prefer the GUA when connecting to the printer.
So when a router withdraws a prefix, it's possible for your computer to still have a (stale) GUA while the printer only has a ULA. Whoops, and your printing is broken.
There is an RFC in review that would fix this: https://www.ietf.org/archive/id/draft-ietf-6man-rfc6724-upda... . It'll likely be approved in 2027, so you'd be able to use ULAs some time in the 2030-s once it trickles down.
15 replies →
The extension headers are fine, because they frequently are not needed, so the fact that they are optional allows the IPv6 header to be significantly simpler than the IPv4 header, which has some seldom used fields which are always present.
No, they are not fine. Their design means that the router needs to know all possible extension headers to just be able to route the packets. Oh, and you also have to do offset-chasing in the hardware fastpath.
As a consequence, IPv6 with extension headers are just dropped by routers in the public Internet.
3 replies →
If it works with mainstream Linux, sounds like an Android problem
Android’s developers have long had their own vision on how v6 should work to the detriment of their own users...
Exhibit A: https://issuetracker.google.com/issues/36949085
1 reply →
That's not a very good example.
1) I can count on one hand the number of times my Internet connection has gone down in the past 5 years.
2) I can count on zero hands the number of times I've tried to print from a phone, ever.
I explained the problem in a sibling post on this thread.
But you want a more complex challenge? Here you go!
You have a small business office that has two Internet uplinks for redundancy. You want to balance the load across them and failover broken links.
This can be done with literally zero setup with IPv4.
9 replies →