My theory is that 5G will be like IPv6. Nobody wanted it, nobody understands it, it makes everything more difficult, and it will take ages to become the standard.
IPv6 makes a great deal of things more easy. It allows smaller routing tables for example. No more NAT makes p2p communication much easier. Yes, addresses are way harder to type now, and that's obviously annoying. But ip addresses weren't made to be typed manually, that's what DNS is for.
> ip addresses weren't made to be typed manually, that's what DNS is for.
I’ve heard this a lot but it doesn’t ring true. I believe I’m in a category with many others where your work involves configuring networks, especially LANs, and you are often entering IP addresses.
I feel this worry about having to type in /128s is overblown. The only times I've had to type a full /128 when setting up my IPv6-only homelab was for adding DHCP static leases for my pet machines.
If you're configuring LANs you're unlikely to be configuring anything deeper than a /64 per LAN, so the effort is approximately the same as IPv4 (four numbers, except that each number is four hex digits instead of three decimal digits).
Similarly, if you're setting up IP rules on a firewall, you're unlikely to care about anything smaller than a /64. If you want to ban a bad actor, blocking a specific /128 isn't going to achieve anything, since the bad actor likely has the ability to use any address within the /64 (SLAAC). You'd just ban the /64.
Lastly, if you're picking your /128s like the static DHCP leases case I mentioned above, nothing prevents you from zeroing all the segments you don't care about. Each of my static leases has all zeroes in the lower /64 except for the last hex digit. Net result is 2001:db8:1234:1::1, 2001:db8:1234:2::1, 2001:db8:1234:3::1, etc. The 2001:db8:1234::/48 is what I get from my ISP so it's already in my muscle memory, so it's negligible extra effort to remember individual machines' IPs.
The OP is probably talking about implementation rather than design, which to this day is very fragile and prone to breakage and misconfiguration, at least on consumer grade networks.
And with certain ISP's being done with DS-Lite which is the worst possible solution at the moment. Sure, your core network is now IPV6, but everyone is going through CGNAT for the parts of the internet that are IPV4 only.
It's one of the reasons I moved from Virgin Media (Cable in the UK) to Zen (FTTP) ... proper dual stack so I have native IPV4 AND IPV6.
> DS-Lite which is the worst possible solution at the moment
No, CGNAT-only is the worst solution.
> everyone is going through CGNAT for the parts of the internet that are IPV4 only.
What choice do they have? There are more humans than IPv4 addresses, so if every ISP were dual-stack, the price would go to infinity. It would be more productive to complain about CGNAT-only ISPs, than those who are actually trying to fix the problem.
