https://www.google.com/intl/en/ipv6/statistics.html ipv6 rollout is well in progress. AWS started charging extra for IPv4 ips. 32 bits is not enough. According to that site India is 70% ipv6
Websites should have the easier time providing ipv6. In the past they could keep it simpler because users weren't on ipv6, & often that was because ISPs have dragged their feet. That's changing
I still wonder why they did not enhance IPv4 by adding more digits. Maybe that would have been easier ?
The phone company did that 50 years ago. When I was a kid we only needed to dial 5 numbers for local, 7 for long distance with a 1. As time went on, additional digits were added.
I mean, that's not too far off from what IPv6 is. The main difference, at least as far as the visual addresses go, is that IPv6 also supports letters, which drastically increases the total number of addresses beyond what extending the number of characters would do.
I think part of the idea was that, since transitioning from IPv4 to IPv6 is going to be difficult, we might as well make it worthwile be ensuring that another transition never needs to happen again for the forseeable future.
there are no "additional bits" left to use and it would also require cosiderable effort to implement
Adding more digits sounds simple, but there is no good room in IPv4 packet. And then any solution using existing one would not anyways be backwards compatible. So for each round of new byte you add to header you would have different version anyway. Requiring new hardware and software.
IPv6 is the only viable path forward. There are no redos here. Saying IPv6 can't work is the equivalent of burying your head in the sand.
IPv4 space has run out. Any solution that add more space requires full adoption from client on the planet.
The thing is, IPv6 only clients can reach IPv4 only services. It's why these services don't advertise v6, they don't technically need to even if some clients would be better off.
But an IPv6 only service _can not_ be accessed by an IPv4 only client.
These limitations are true for any possible alternative to IPv6.
We need people to stop trying to deny IPv6, it IS the future.
If the USB consortium can do it, the network folks can as well.
Implementing it fully is a technical process, but it's not an impossible barrier. The real reason for the slowness is that there's little to no profit incentive to ensuring its full compatibility, in part because IPv4 is guaranteed to be (mostly) backwards compatible for decades.
It's a slow process, but it's happening. There's no need to invent yet another standard for something that's already working and will likely be good enough to last generations once it's finished being implemented.
Nothing is going to change the fact that _any_ solution needs every router to work, needs to handle legacy v4 and new v* addresses.
There is a reason why there aren't even alternative proposals to IPv6, none of the problems that the IPv6 deployment is having can be solved by a "simpler" implementation.
But that's just not true now, is it? Extending the address space didn't need a complete rework and rethink how local network routing works, removal of standard NAT and messing around with address assignment (DHCPv6, SLAAC, and all of those messes).
IPv6 isn't having problems because it's not compatible, it's having problems because they made an utter mess of how local networks works and routers most companies (and people) have aren't able to reliably deal with the complexity of what's rather easily solved by DHCPv4 on "old" networks. The fact that IPv6 decided to rework the basics of how routing works while also demanding a dual-stack deployment is the core of the problem.
If the extent of deploying IPv6 would be "we'll enable address assignment on our existing DHCP server with same rules", you'd be on mostly IPv6 right now.
I think the issue here is that IPv6, despite being the successor to IPv4, is actually quite old in itself; it was developed starting in the early 90's and so it's just a different idea of what the internet was going to be. NAT is more like a "hack" that developed around that time as well and has turned into the normal way networks are managed.
(But that was even shown when they were dreaming of deploying IPv6 without any kind of DHCP which had to be backtracked making a utter frigging mess of address assignment on networks with multiple competing approaches.)
Who? And what constitutes "actual"? Are the netengs in FAANG actual network operators? Who aren't actual operators? Why they aren't?
>Most deployments of IPv6 are mobile networks which don't actually allow their client phone IPv6 addresses to be reachable (as was the dream around getting rid of NAT).
[Citation needed]
I just tested with 2 cellular ISPs, and I was able to connect to the SSH service on my phone with either.
(It's mostly useless though, since whenever the phone is locked the app is also suspended, causing the SSH service to be dropped)
There's no fundamental difference in routing. They both use trivially simple prefix routing.
> removal of standard NAT
NAT isn't "standard". NAT is a nasty kludge that people put in because they were too lazy to upgrade their software stacks. And IPv6 was defined before NAT was commonly used, in large part as a way to avoid NAT becoming common. Because NAT is a bad idea in every possible way.
Also, IPv4 is itself much, much simpler without NAT. And you can do NAT on IPv6 if you're stupid enough to want to.
Saying that IPv6 is complicated because it removes the need for a complicated hack like NAT is just insanity.
> (DHCPv6, SLAAC, and all of those messes)
Those are not in any way more complicated or harder to use than DHCP4. Setting up a DHCP server to get automatic address assignment on an IPv4 network is actually harder than letting a router automatically do SLAAC to do address assignment on IPv6.
Also, DHCP is not part of IPv4, and DHCP6 is not part of IPv6. I was using and managing IPv4 networks before DHCP existed. Yes, it was a royal pain in the ass... which is why that functionality was integrated into IPv6. Which was done a protocol that's actually simpler than DHCP, although I didn't and still don't approve of the way it wastes address bits.
The fact that you consider SLAAC somehow equivalent to DHCP shows that - I've never seen a router (outside maybe some ciscos) where those two could be used to achieve the same stable network assignment and configure downstream service firewalls appropriately. That is - if SLAAC actually works across all clients and they're not a buggy mess.
45% adoption.
>horrid bugginess of IPv6 network stack
Were the IPv4 network stack _not_ buggy? Are you sure it's just not the general incompetence of network equipment people?
>where those two could be used to achieve the same stable network assignment and configure downstream service firewalls appropriately
What?
>SLAAC actually works across all clients and they're not a buggy mess
What buggy mess? I'm on SLAAC and I don't notice anything buggy?
After 28 years? Yeah, that sounds like failing adoption.
Edit: I should caveat - I don't have good data on actual lifespan of networking equipment; if you can show that it's likely to live more than a decade then that would be reasonably compelling.
I stand by my assessment that 45% adoption after 12 years is actually pretty good.
[0]: https://transparencyreport.google.com/https/overview?hl=en
(I use Google's data for both HTTPS and IPv6 stats. If you consider 3rd party data, HTTPS adoption is way worse, even to this day it hasn't topped 75%.)
This is a classic responsibility-evasion cycle:
Users: We can't get equipment that supports this (or supports it well, or makes it easy), so we won't deploy it.
Vendors: Users aren't demanding it, so we won't support it.
The result is a suboptimal local equilibrium and an obvious market failure, but you invariably get loudmouths on message boards saying "Look, the Market (Peace Be Upon It) has spoken!". Gets a bit old.
The right answer in this case is probably coercive government regulation of the form "No IPv4 packet shall cross a property line", but I'm not holding my breath.
You can scream and cry as muck as you like, IPv4 IS the present. And people in general will not care until the world will work for them.
And I trust those companies to know how to cater for people's money. When, and if, IPv6 will be the right tool, they will use it. But trying to shame them on such a meaningless technicality is the equivalent of criticising NASA for the colour of the Apollo launch station.
IPv4's address space is fundamentally too small, so there has to be some kind of alternative or replacement.