NAT64/DNS64 (iOS): Your device only gets an IPv6 address, but when it queries an ipv4onlydomain.com, it receives an IPv6 address mapped to the domains IPv4 address. When it sends packets to it, they are translated by the ISP to IPv4. All their customers remain IPv6-only. This method requires apps to be able to handle IPv6 addresses, which Apple enforced.
XLAT464 (Android, Windows): your device gets an IPv6 address. The OS translates IPv4 packets to a special range of IPv6 before sending them out, because you’re only connected over IPv6. The ISP translates them to IPv4 to reach legacy services.
The benfit is: absolutely no NAT for IPv6 packets. Complete end-to-end connectivity.
192.168.0.1 shouldn’t be glorified. Good riddance. We have mDNS now, just have the user type router.local into their browser.
Fun trivia:
A few years ago I wrote a kernel module, that together with existing kernel mechanisms allowed almost all transition tech at a time, and ended up in a bunch of CPEs:
https://github.com/ayourtch/nat46/tree/master/nat46/modules
You can see the pull request activity... two pull requests in the past month... before that it was last time in 2017. So some things seem to be moving.
do the privacy extensions (which I've heard of, but don't get at even a basic level) allow a little privacy in this scenario?
Remote servers will see the privacy address, but they won't know which machine the address belongs to. The addresses will also be abandoned after a maximum of a week, so you can't trawling through your old server logs to find active v6 addresses either.
They'll still be able to see what network you're coming from, but that's not something that NATing outbound connections from your network would help with anyway.
Well...
Comcast made IPv6 on-by-default to all residential customers some time ago (IIRC) and available (not sure if on-by-default) to all business customers. [1] Verizon requires all LTE devices to support v6 [2] and achieved >70% penetration in 2016 [3] (I'm too lazy to find a more recent statistic.) T-Mobile launched v6only in 2014 [4] and hit >90% in 2018. [5]
> Eventually, of course, the dam is going to break and IPv6 will become ubiquitous. ISPs will decide that buying blocks of IPv4 addresses costs more than providing new customers with IPv6 plus some kind of carrier-grade NAT for legacy IPv4 addresses. More website hosting companies will support IPv6 in response, and suddenly IPv4 will be so last-decade.
In short, that happened.
[1] https://business.comcast.com/help-and-support/internet/comca... [2] https://www.apnic.net/wp-content/uploads/2017/01/vzw_apnic_1... [3] https://archive.today/20160719154102/http://www.worldipv6lau... [4] https://www.internetsociety.org/resources/deploy360/2014/cas... [5] https://pc.nanog.org/static/published/meetings/NANOG73/1645/...
This implies an ISP can do math. Frontier acquired lots of IPv4 addresses w/ it's $10B purchase of Verizon assets(in 2015 & $7B in 2010 & 2B of AT&T).
Frontier has no plans to ever deploy IPv6.
According to https://www.google.com/intl/en/ipv6/statistics.html#tab=per-..., IPv6 adoption in the US is above 42%. Was that all done by lunatics?
I occasionally find websites that are broken and don't load, but that only happens rarely and it's not exactly a problem that's exclusive to v6.
if you disagree, i'd appreciate if you could explain why...
> they only run into problems if they connect to a wifi that doesn't support IPv6
is like 99% of the problems IPv6 devices encounter.
at least it's a solveable problem, because it's in the customers sphere of influence, whereas getting all websites i am interested in onto IPv6 is not.
em-bee: IPV6 IoT only run into problems if X
avianlyric: X is like 99% of the problems IPv6 devices encounter.
em-bee: thanks. i didn't realize it would be that bad.
What new information did avianlyric bring? Was that not a re-statement of what came before?
only running into problems if X
implies that X is a small problem.
but 99% points out that X is a big problem.
By all accounts, many of us in the networking space predicted that this would absolutely, positively, without a doubt occur by 2010.
Just keep shipping routers that support both IPv4 and IPv6 and you won't have a problem. If IPv6 gets popular enough, you won't even need NAT on that interface--it will just be an admin network. This is only an issue when major OSes start dropping IPv4.
See also fossil fuels.
The problem with ip6, if there was only just 1, is that they cared more about the engineering than what customers wanted/needed. their attitude was: if they dont use it, well they are just stupid, everyone else will use it and have to switch anyways.
From an ISP's point of view IPv6 is a lot of No Business Case. Those three words are the death knell for any proposal to do anything in any business that expects to be here next year. It has exactly nothing to do with the geeky I-could-have-designed-it-better arguments about the technology.
If you are an ISP that is a going concern, with a bunch of customers sitting on IPv4 addresses, then handing out IPv6 addresses makes no difference, except when it breaks something. You still have to give your customers just as many IPv4 addresses. So why bother?
If you plan to migrate your customers to IPv6 then you are a lunatic. Its going to break stuff. Lots of websites don't exist on IPv6, and customers are going to notice. Also your customers have spent the last 20 years slowly picking up bits of IPv4 lore, like the vital importance of 192.168.0.1, and are going to be puzzled when this doesn't work any more. All of this translates into higher support costs and more customer churn.
Also, your allocated block of IPv4 addresses is a valuable asset in its own right; not only does it have real financial value (around $20 per address at present), but it also acts as a barrier to entry for competitors; if you want to set up in business as an ISP you are going to have to acquire some IPv4 address blocks from somewhere, and they aren't making any more of them. Managers are trained to look for barriers to entry to their industry, and the current IPv4 situation is exactly that. No sane manager is going to do something to make it easier for new competitors.
Eventually, of course, the dam is going to break and IPv6 will become ubiquitous. ISPs will decide that buying blocks of IPv4 addresses costs more than providing new customers with IPv6 plus some kind of carrier-grade NAT for legacy IPv4 addresses. More website hosting companies will support IPv6 in response, and suddenly IPv4 will be so last-decade.