Are we past peak IPv4?
sidn.nl
sidn.nl
It's hard to estimate how increasing percentage of users with IPv6 access is impacting the demand for public IPv4 addresses. My instinct is "not by a lot", but it's interesting to think about.
Currently IPv6 does not allow you to simply do away with public IPv4 adderesses. On the ISP side that would mean providing a product that couldn't access a huge chunk of the web. On the hosting side that would mean providing a service that isn't available to a huge chunk of users. It's not viable.
But increasing IPv6 deployment could possibly reduce IPv4 demand. If you're an ISP and half of your customers' traffic is over IPv6, maybe you could densify your carrier-grade NAT setup, putting more users behind the same number of public IPs. If you're a large content provider and half of your users are over IPv6, maybe only half of your load balancers need public IPv4 addresses.
The question is - is any ISP or provider actually doing those things in response to increased IPv6 traffic?
> It's hard to estimate how increasing percentage of users with IPv6 access is impacting the demand for public IPv4 addresses.
I'm more interested in this excellent point you make, about the economics. I don't think this has been thought through as well as the technology.
And can we please find additional sources other than GOOG for IPv6 adoption?
I've definitely seen this. That's how IPv4 connectivity became an enterprise/business line feature on some ISPs.
A neglected detail about CGNAT is that it has lower the quality of IPv4 connectivity. Higher latency + more connectivity issues.
So if you offer a service then you should offer it on ipv6 due to that.
That's a really slow transition though. So I wouldn't expect this to kill off IPv4 any time soon (say 10y), but longterm it's been the trajectory already.
If most if not all services are available also on IPv6, the demand for IPv4 may go down.
Services are the most forced side for using IPv4. Consumers can be behind some sort of NAT, so you can have a lot of users behind single IP addresses. But the destination, if it must be IPv4, is not so trivial.
That is the low hanging fruit of IPv6 adoption, having for all interesting services IPv6. Then ISPs will be free to do at their own rhythm their migration to v6 only. If interesting services start to be only for v6 (because they decided that public adoption was high enough, at least for their main target markets) ISPs will start to add ipv6 to their users, or provide some of the transition mechanisms like NAT46 if that is too complex/expensive because local infrastructure.
But still a lot of very used services are IPv4 only (https://whynoipv6.com/ have a recent enough list ), or worse, have faulty IPv6 implemented (i.e. AAAA records but nothing listening there or with routing problems)
You make a great point, but I would stress that the problem is not posed by IPv6 but by major ISPs with a global presence, such as Vodafone, which unexplainably have been dragging their feet on IPv6 support.
The biggest telco (Telstra) in my country (Australia) allocates IPv6 addresses (only) to mobile phones. I'm not sure when the change over happened, because no one noticed / mentioned it. The WiFi hotspot on my WiFi phone hands out both IPv4 and IPv6 addresses. Each host gets it's own /64 routable IPv6, so there is no NAT on IPv6. By Debian laptop prefers to use IPv6 when it's available, so when hotspotting to my phone I'm entirely IPv6.
I'm not sure when that happened, because I didn't notice. As far as I can tell, no one noticed.
In my ISP/cloud provider days sure I spent a lot of time implementing v6. But it's just nowhere near worth the cost to mess with particularly when you still want/need public v4 along side.
Also, I don't understand how v6 could be responsible for these post peak prices. Most everyone still has the v4 addresses as well and no way v6 only cowboys are outpacing general growth.
My company has bought other companies in the past and inevitably we've had IP space conflicts. Work arounds involved requiring users to connect to different VPNs to use services running in different networks. The end result was that we had to renumber huge chunks of the company and patch up the routing tables to make this work.
If we'd both been running IPv6 there wouldn't have been any IP space collision. We could just connect them together and be done with it.
The way you structured it, I feel like the analogy for "living on Mars" would have to be "merging company networks", but that happens all the time.
If Mars is supposed to be IPv6 then I can't figure out what the utilities are supposed to represent.
I was making a "joke" that it's as likely as us getting to another planet and it being already setup to be habitable by humans.
It also only really works when it's a 1:1 scale, as you add more networks the configuration grows with N^2.
For this to work you need to expose the services to those non-overlapping blocks, ie write DNS views and associated shenanigans.
At this point you are half-way to IPv6, so why bother with implementing NAT44 at all?
Maybe I have a significantly different setup, but on the SaaS side this only requires binding to to ipv6 and adding an AAA record on your ingress usually.