> I'm one of their customers. It's pretty good - their provided router is locked down to hell and they're on a cgnat
This sounds like mine. I'm guessing yours doesn't support IPv6 because most fiber providers don't.
For the router, I already build firewalls so that. I pay $10/mo to escape their cgnat.
I've also alerted them to expect regular haranguing from me about deploying IPv6. Especially since bgp.he.net shows they have a /40 allocated to themselves; it doesn't seem to be used.
I've had less than 0.5% of customers ask for IPv6 from my fibre ISP. It's not worth supporting as a result. The main reason is that any service that is not widely used will have gremlins that result in poor customer experience, and if it's always the same handful of customers hitting problems or finding quirks, there is a real risk of poor word of mouth incident reporting that can harm the business. At least if something goes wrong with IPv4, it's going to be noticed very quickly.
Some people will say monitoring is all that you need, but I do not agree. There are a million different little issues that can and do occur on physical networks in the real world, and there's no way monitoring will have a 99% chance of detecting all of them. When incidents like the partial Microsoft network outage that hit certain peering points occurred, I had to route around the damage by tweaking route filtering on the core routers to prefer a transit connection that worked over the lower cost peering point. It's that kind of oddball issue that active users catch and report which does not happen for barely used services like IPv6.
> I've had less than 0.5% of customers ask for IPv6 from my fibre ISP
How many ask for IPv4? I understand your situation, it's a lot of work, for something that many won't notice. It's just that saying there's no demand because your average consumer, who also doesn't know what IPv4 is, isn't asking for it, is the mentality that keeps IPv6 from being implemented.
On the funnier side of things, we've also sometimes run into the opposite problem that we can't reproduce an issue, because it's only on IPv4 and 95% of the time everything we do is IPv6. But we're also not serving home users.
Static IPv4 addresses are closer to around 5% of customers. Nobody asks for IPv4, but some customers bring their existing or own wireless routers along and occasionally choose devices that are not IPv6 capable. Maybe in another 10 years those devices will finally be fully removed from service. The worst stragglers right now are the old combo DSL modems that effectively have no modern replacements -- it's just not worth spending money to replace them when customers are going to migrate to fibre soon enough.
I don't think GP meant static addressing but literally, how many ask for IPv4 service? None, because you just provide it; it's an unstated expectation.
Now apply that to IPv6 and you can see the point that (I think) GP is making.
Side note: The claim it is not widely used doesn't track. How many people use Google or Facebook? More than half of that traffic is over IPv6.
https://circleid.com/posts/ipv6-usage-in-the-u.s-surpasses-5...
So when you said `ask for IPv6` you meant `ask for a static IPv6 prefix` or something else similar to a static IPv4 address? Or is this an apples to oranges comparison?
And then you say `Nobody asks for IPv4` - so nobody asks for IPv4 and 0.5% ask for IPv6?
> I've had less than 0.5% of customers ask for IPv6 from my fibre ISP. It's not worth supporting as a result.
Big, evil, hated Comcast has full ipv6, and I doubt any of its customers asked for it either. Instead people complain they’re only getting a /60.
Comcast was forced to go to IPv6 because they ran out of IPv4 addresses in the private address space to use for management of their network (think of how each and every cable modem needs a an address for management in addition to all the routers and CMTSes). I was a fly on the way inside one of the router vendors when this took place more than 15 years ago.
If you already have to do CGNAT, why not IPv6 as your core network with NAT64 at the border and 464XLAT on the CPE? It gives you best of both worlds.
I'm not doing CGNAT. We were able to get enough IPv4 addresses directly from ARIN a few years ago after being on the waiting list for a couple of years. It's a pity that widespread fraud depleted that pool faster than it should have been.
CPE support for IPv6 has generally been garbage with it taking 15-20 years before the bare minimum was supported by mainstream router vendors. Even today there are still vendors that assume only IPv4 support. In my opinion the IETF really screwed up when they made IPv6 more complicated than just IPv4 with more address bits. The incumbent in my area generally uses PPPoE in their access network, but routers that supported PPPoE and prefix delegation basically didn't exist in 2010, and only started being available circa 2015 (in part due to the required bits not existing in OpenWRT and the hardware vendors' software development kits for their chipsets). Sure, we're 10 years further on now, but there remain a number of vendors that only support IPv4 for management of devices (cough Ubiquiti cough) in parts of their product line.
That said, there are features of IPv6 that are absolutely awesome for carriers. The next header feature that pretty much eliminates the need for MPLS in an IPv6 transport network is one such item that makes building transport networks so much cleaner when using IPv6 than IPv4. No more header insertion or rewriting, just update one field and fix up the delta on the checksum and CRC. They just aren't really applicable for smaller networks.
Ah, okay, well if you already have the IPv4 address space available for all your customers that's a different story. I can understand why you want to wait to dual-stack in that case.
I do think NAT64/464XLAT is a pretty good architecture for new ISPs that can't get their hands on IPv4 space, though. Or even MAP-T, but CPE support isn't really there yet.
Android and iOS have pressure on app developers to support IPv6 or at the very least function on 464XLAT. On home broadband people could connect anything including programs that use literal IPv4 addresses and break on 464XLAT. Things like corporate VPNs, Skype, online gaming. Ironically these services would benefit the most from supporting IPv6 but they've evolved to deal with IPv4 NAT to the extent that they've become dependent on IPv4 connectivity.
For me, no IPv6 = no business. I don't think it's acceptable to build a network on IPv4 only at this point, it speaks to being willing to cut corners and not do things the right way just because it's easier.
I worked in this space for a while, in the US. Outside of the major cities, Internet service falls off extremely quickly. Like, shockingly so: you can be as close as fifty miles from, say Philly or Flagstaff and have zero fiber, zero cell coverage, just nothing.
The people who attempt to fill these gaps are commonly rural telephone companies, electric cooperatives, tribal entities, or mom and pop shops where the owner grew up on a Ditch Witch and only knows as much IP networking as essential to light up the fiber and get the packets flowing upstream.
They are enormously resource constrained in ways you might not expect, too, eg operations can grind to a halt because everyone is out with a chainsaw after a storm, or because the Guy that Knew Stuff about their network died suddenly.
They are very, very unlikely to decide to run an IPv6 network just because. There's no upside that makes the juice worth the squeeze for them.
In principle, IPv6 core networks can actually be very beneficial for small providers just starting out if they're not able to get IPv4 addresses for their customers and are forced to use CGNAT.
In an IPv4-only CGNAT setup, all the traffic has to flow through the CGNAT gateways, and that gear is stupidly expensive. Having IPv6 in the mix means that anything that supports IPv6 (such as most streaming services) won't hit the CGNAT gateway and can just be routed natively. This can really save money on CGNAT hardware.
For implementation, you can use NAT64/DNS64 for your CGNAT setup and implement 464XLAT on the CPE. This keeps your whole edge network IPv6-only so you don't have the complication of maintaining two parallel configurations on the edge.
There is also MAP-T, which is even lighter on infrastructure since it pushes all the state into the CPE and avoids the complication of stateful CGNAT. But unfortunately CPE support for it is pretty limited at the moment.
> Outside of the major cities, Internet service falls off extremely quickly.
Saline is less than 10 miles from Ann Arbor.
> The people who attempt to fill these gaps are commonly rural telephone companies, electric cooperatives, tribal entities, or mom and pop shops...
That's fair, but at some point, you need to recognize you are competing with a major ISP. No one appreciates it when you come in, tear up the roads, and then pull out once the incumbent ISPs bump up their speeds ever so slightly. (Looking at you, Google.)
> They are very, very unlikely to decide to run an IPv6 network just because.
No one deploys IPv6 "just because," and yet more than half of the traffic to major sites is IPv6.
> Outside of the major cities, Internet service falls off extremely quickly
I live half an hour from a state capital and my only option is cable... the coaxial cable they laid bare on my flower bed decades ago. I dig it up about every other year when planting. It's not even in a conduit!
I wish I could say no IPv6 no business. There are only 2 ISPs here, one cable and one fiber. Neither have IPv6, the smaller ISP also does CGNAT because IPs are expensive. I'm trying to convince them that they could save money with less powerful CGNAT hardware if they deploy dual stack.
I agree in principal but if the only other option is Charter/Spectrum/Comcast, you bet I'm going with the "lazy" person's fiber.
I have spent most of my career under the thumb of fucking cable and I'd sooner slam a car door on my nuts than go back to paying so much money for such garbage service.
"I'm guessing yours doesn't support IPv6 because most fiber providers don't."
Yeah, what's up with that? I just got switched on to fiber and the CGNAT for IPv4 doesn't shock me much, but what's with the no IPv6 in 2025?
I know enough to deal with it, but what's the deal? Is there something systematic here?
Everybody can muddle along without IPv6, so it's easy to make it a very low priority. Especially for small shops that are struggling just to create a viable business. IPv6 needs something more to motivate it, a web destination or application that is only available on IPv6.
We used to have freeipv6porn.com, lol. But I suspect that was a joke as much as anything else given how much porn you can get for free all over the Internet.
The eggs need some chickens first.
> IPv6 needs something more to motivate it, a web destination or application that is only available on IPv6.
How about not having to pay for (as) beefy CG-NAT hardware because people that go to Youtube, Netflix, MetaFace, TikTok, etc, can directly connect via IPv6.
Hadn't thought of that, but it might not be a huge savings unless you were to go ipv6 only. If you're still going to support ipv4 anyway, the hardware savings might not be too significant.
> Hadn't thought of that, but it might not be a huge savings unless you were to go ipv6 only.
Even a small number of devices/services not supporting IPv6 can have huge costs:
> Our [American Indian] tribal network started out IPv6, but soon learned we had to somehow support IPv4 only traffic. It took almost 11 months in order to get a small amount of IPv4 addresses allocated for this use. In fact there were only enough addresses to cover maybe 1% of population. So we were forced to create a very expensive proxy/translation server in order to support this traffic.
> We learned a very expensive lesson. 71% of the IPv4 traffic we were supporting was from ROKU devices. 9% coming from DishNetwork & DirectTV satellite tuners, 11% from HomeSecurity cameras and systems, and remaining 9% we replaced extremely outdated Point of Sale(POS) equipment. So we cut ROKU some slack three years ago by spending a little over $300k just to support their devices.
* https://community.roku.com/t5/Features-settings-updates/It-s...
* Discussion: https://news.ycombinator.com/item?id=35047624
Sadly the post is now behind a login: what happened later was Apple donate a bunch of Apple TV devices to the tribal ISP and that cut their IPv4 usage by an order of magnitude (or some ridiculous number) and there were major savings. The ISP then recommended AppleTV to all of their customers to get the best viewing experience (because of the latency/overhead of CG-NAT when streaming video).
So the more you move over the more headroom you have for the broken IPv4-only systems. AIUI, the rollout of MAP-T/E has helped in that things are more stateless, and more work is done at the CPE, but there's still overhead.
Thankfully, they are doing IPv6, although one day I had some weird issue where IPv6 was broken but if I disabled it ipv4 was still working. Could have been my fault, IPv6 is generally new to me (not much of a network person).
I get the impression that they are still learning to run an ISP, both technically and customer facingly. It's weird - I learned more about them from this article than from actually being living here with them.
Surprised they aren't deploying NAT64/DNS64 with 464XLAT on the CPE. You get essentially the same setup as CGNAT for IPv4 services but your whole core network is native IPv6 so you only have one set of address space to manage and your customers will be able to directly connect to anything IPv6 related.
How would you as a customer tell if they were?
Because you'd have native IPv6
Comcast has pretty good IPv6 support
since tailscale exists, why would you care about cgnat or even pay to escape it?
I'm not the only person connecting to my machines.
Some applications want to open ports and don't have the server-side infrastructure to punch a hole through NAT. Especially P2P apps and some games.
Sometimes I want to run a small, low-traffic web server from home.
Sometimes I'm connecting to my network from a machine that I don't control and can't install Tailscale on.
Tailscale uses the same range as CGNAT.
https://github.com/tailscale/tailscale/issues/12829
You shouldn't be seeing the CGNAT addresses inside your home network, should you?