Ask HN: What we as community can do to adopt IPv6 faster?
All I can think of is tweeting, emailing big corps or maybe creating a petition, so that these companies take this seriously. Anything else that you guys think maybe more effective?
There’s no need for it, apparently.
The next internet protocol should have stronger arguments, and perhaps start with being interoperable with IPv4.
I'd focus on simplification and killer features.
Are there any sites that have fully migrated to ipv6?
The sites that I’ve seen running v6 at all are doing so in parallel with v4, which I think just underscores your point.
Is there any argument that's stronger than having ran out of IP addresses?
We already have service providers charging a premium for ipv4 addresses. Isn't that a good enough incentive?
I personally would rather pay a cheap premium for a few ipv4 my company needs rather than having to migrate to ipv6. I would most likely still need a few ipv4 anyway.
It really doesn't look like it.
IANA already announced in 2011 that the free pool of addresses was depleted.
Europe's regional internet registrou already announce two years ago that they already allocated their last remaining address pool, and now all they can do is manage a waiting list where they assign addresses of companies which went out of business.
https://en.wikipedia.org/wiki/IPv4_address_exhaustion
To further refute your point, here's an announcement from Hetzner stating that ipv4 prices skyrocketed due to the exhaustion of ipv4 addresses.
IPv6 definitely has its issues. But I recommend you have a look at these graphs:
However you will always run into the exact same issues we are facing with IPv6 today.
There are a whole raft of advantages:
1. NAT/CGNAT[0] becomes irrelevant (although you can still use local/non-routable prefixes if you wish), eliminating many of the issues around port forwarding and reachability of services;
2. While IPSec is broadly available for IPv4, encryption and authentication are baked into IPv6;
3. The enormous size of the IPv6 addresses would allow every location to have their own 16 bits (or more, if necessary) of routable address space;
4. Despite NAT, we have already run out of IPv4 address space[1] across all regions.
There are other good reasons for IPv6 too, but those are the most important IMHO.
- You get extra bits to play with for addressing. Fly.io has done some cool stuff with their "6PN" addresses, built on top of IPv6 and Wireguard. They encode information in the address itself and use eBPF to route on it.
- As a result of everyone essentially having infinite public addresses to assign for themselves, we can get rid of NAT, which would enable peer-to-peer apps. See Tailscale's "How NAT Traversal Works" post for a great survey of the techniques used to connect devices behind NATs. (Note that getting rid of NATs does not mean getting rid of private networks or block-all-inbound firewalls. IPv6 allows both of these security measures.)
- Some more points here [2] I hadn't seen, particularly around better routing efficiency by getting rid of IP checksums and allowing for smaller routing tables.
[0] https://ipv4marketgroup.com/ipv4-pricing/
[1] https://tailscale.com/blog/how-nat-traversal-works/
[2] https://www.networkcomputing.com/networking/six-benefits-ipv...
IPv6 adoption is going to drag along, unless there is a compelling reason for people at large to adopt it.
There are reasons, but most of them don't apply to companies that either have enough v4 space or have enough budget to get the space they need.
As more consumer networks push towards v6 first/better, you'll see more hosting support v6 as an equal option rather than a second class option, but I don't think we're there yet. A lot of LTE networks do v6 as well or better than v4, but a lot of residential networks still don't do v6 or do it half-ass. A host doesn't lose a lot by being v4only, dualstack adds extra complexity, and v6only loses a lot of potential customers.
We don't need it because we can NAT.
We don't want it because we prefer privacy, not being easily tracked all over. Router walls over firewalls.
Multicasting didn't work out neither. No common ISP invested into more RAM for their routers just to stream a TV show more efficiently.
QoS contradicts the basic internet policies.
Personally I think the only realistic move is to advocate for an IPv7, which is backwards-compatible with IPv4 (keep ARP, keep classless addressing, keep DHCPv4, keep seeing NAT as acceptable, don't rely on multicast, etc.) instead of the wacky re-engineering project that IPv6 is. There's a clear need to get more addresses. There's not a clear need for all the other stuff being pushed as part of IPv6.