3 ms·

It would be interesting to know how many Internet users do not have IPv4 these days...

zamadatix 4y ago
Wired very few and unlikely to change quickly, mobile a good number and often changing without people knowing it happened. Mobile has a lot of NAT464 where the middle layer is the carrier and all v4 gets translated over v6 which is all that's assigned to the user and then out a gateway service. In wired you'd need to insure everyone had a device which understand to do that (CLAT in 464XLAT jargon) so carriers have been doing NAT444 where there are 2 layers of v4 NAT (home users then carriers) before the carrier's core does the final translation to a real v4. NAT444 is clunky and doesn't even help make sure users get to v6 but it's attractive to carriers because it works without having to have anything change from the customers perspective (beyond their external NAT IP being a 100.64.0.0/10 address instead of a public IP). Because of this the tail end of those who truly do not have any v4 on wired is going to be quite high.
- BlueTemplar 4y ago
  I'm mostly thinking of all those landline ISPs in Asia that I hear never got any IPv4 IPs to start with..?
  - zamadatix 4y ago
    They got IPs but they missed out on the "Hi, I'd like 16.7 million IPs please" era and mostly on the "Hi, I'd like yet another 65k please" era as well. With NAT444 carriers can use a 1 public IP to serve all of the connections for somewhere between 100-1000 customers depending on the usage pattern of those customers. Right now an IPv4 address sells for about $40 so even on the low scale if they had 0 IP addresses that's a one time cost of less than 50 cents per customer. Without NAT444 it would have been 100-1000 times more expensive for newer ISPs to use IPv4 and you would have seen a lot more by now.
bombcar 4y ago
This is the real key - until there are IPv6 sites/users that cannot access IPv4 we won't really see a push to move over.
- pantalaimon 4y ago
  There already are IPv6 only servers and sensor networks. On client connections most ISPs use DSLite these days.
- WorldMaker 4y ago
  There's been some push from the other side: cell phones move too much in physical space to make IPv4 routing efficient (especially with today's scarcity) so today most cellular networks are IPv6 "native" and rely on IPv4-over-IPv6 router proxies for IPv4-only traffic. Smartphones have been among the biggest leading adopters of IPv6 out of necessity to the network topology.

Wired very few and unlikely to change quickly, mobile a good number and often changing without people knowing it happened. Mobile has a lot of NAT464 where the middle layer is the carrier and all v4 gets translated over v6 which is all that's assigned to the user and then out a gateway service. In wired you'd need to insure everyone had a device which understand to do that (CLAT in 464XLAT jargon) so carriers have been doing NAT444 where there are 2 layers of v4 NAT (home users then carriers) before the carrier's core does the final translation to a real v4. NAT444 is clunky and doesn't even help make sure users get to v6 but it's attractive to carriers because it works without having to have anything change from the customers perspective (beyond their external NAT IP being a 100.64.0.0/10 address instead of a public IP). Because of this the tail end of those who truly do not have any v4 on wired is going to be quite high.

BlueTemplar 4y ago
I'm mostly thinking of all those landline ISPs in Asia that I hear never got any IPv4 IPs to start with..?
- zamadatix 4y ago
  They got IPs but they missed out on the "Hi, I'd like 16.7 million IPs please" era and mostly on the "Hi, I'd like yet another 65k please" era as well. With NAT444 carriers can use a 1 public IP to serve all of the connections for somewhere between 100-1000 customers depending on the usage pattern of those customers. Right now an IPv4 address sells for about $40 so even on the low scale if they had 0 IP addresses that's a one time cost of less than 50 cents per customer. Without NAT444 it would have been 100-1000 times more expensive for newer ISPs to use IPv4 and you would have seen a lot more by now.

bombcar 4y ago

This is the real key - until there are IPv6 sites/users that cannot access IPv4 we won't really see a push to move over.

pantalaimon 4y ago
There already are IPv6 only servers and sensor networks. On client connections most ISPs use DSLite these days.
WorldMaker 4y ago
There's been some push from the other side: cell phones move too much in physical space to make IPv4 routing efficient (especially with today's scarcity) so today most cellular networks are IPv6 "native" and rely on IPv4-over-IPv6 router proxies for IPv4-only traffic. Smartphones have been among the biggest leading adopters of IPv6 out of necessity to the network topology.