Hacker News new | ask | show | jobs
by tthebst 2143 days ago
Why are is there such a difference between countries?
2 comments

A large chunk of the US adoption will have been driven by Comcast. They were early adopters and innovators, not least because they have more CPE (cable modems) than are addressable with RFC1918 addresses.
Could be ISP adoption. Some ISPs have been more progressive in ipv6 than others.
My ISP in the Netherlands (Ziggo) provides me with native IPv6 if I use their supplied router, but forces me onto IPv4 when I use theirs in bridge mode in conjunction with my own router.

Still not sure why they do that.

These ISPs are using DS-Lite, Dual-Stack Lite.

https://www.juniper.net/documentation/en_US/junos/topics/top... (this isn't a purely Juniper thing, but they have nice diagrams on their documentation)

It's a kind of carrier grade NAT with 4over6 baked in.

Depending on the version of this they are relying on your modem to perform encap/decap of 4to6, hence when you switch to modem mode or your own router you fall back to what the network truly is... v4.

This is what the knuckle draggers at Virgin Media are contemplating apparently.

In the UK the best option for IPv6 is https://www.aa.net.uk/ but unfortunately for me the DSL speed in my area is pretty bad due to being a few KMs from the exchange.

The alternatives to all of this is to run your own Wireguard instance elsewhere on a v6 network, and tunnel the entire home network to it.

DS-lite gives you a v6-only internet connection. v4 is provisioned as a service over the top of that, using a tunnel between your router and a server inside the ISP. The underlying network is v6, so a router without DS-lite support will only get v6 (which will generate support calls because "your router must support DS-lite" is too complicated for many people to understand).

My guess is that turning on bridge mode also migrates you from the ISP's newer DS-lite service to their older v4-only one. This is unfortunately common in DS-lite deployments; ideally the old service would also have v6 so that you aren't forced to choose between v6 and non-CGNATed v4.

> unfortunately for me the DSL speed in my area is pretty bad due to being a few KMs from the exchange.

For VDSL the distance to an exchange shouldn't matter. The copper runs to a green box in the street, then another newer-looking green box nearby (or in some cases attached) with fibre in it takes the VDSL signal - and only old-school voice line conversations go to the exchange.

So definitely if previous ADSL speeds were poor it's time to check again. If you just meant that as shorthand then no worries, but I've run into way too many people who have the idea that all DSL is the same.

IDnet does IPv6 and give you a /48.

I was on them for a year great service.

I'm now on Hyperoptic which also does IPv6.

BT and Hyperoptic both also provide native IPv6 in the UK

Hyperoptic will give you a globally unique IPv4 address for an extra £5/month, and otherwise will stick you behind their CGN.

I have the same situation with Vodafone Cable in Germany. I was told it is because of 6-to-4 bridging. A customer on IPv6 must be able to bridge to IPv4 to reach IPv4-only servers, but a customer on IPv4 will probably never encounter IPv6-only servers (except when using an IPv6 testing tool).

I'm not a networking expert, but AFAIK there's nothing stopping them from making their 6-to-4 bridging available to customers with a custom router. However, getting that stuff up and running correctly can be tricky, so it would involve a lot of support burden on the ISP's side. Hence why they don't offer that, they just drop those customers down to a configuration where the customer gets to use well-known protocols only (IPv4 and DHCPv4 over Ethernet).

I'd assume it's because they expect that if you are using your own equipment it could be older and by just forcing v4 it reduces support costs. 100% pure speculation.
In Australia I get a /56 block
I think it's often out of necessity, when you have rapidly increasing numbers of devices (smartphones, IoT) and a large population. This could explain why IPv6 deployment is widespread in Asia. At some point maintaining huge CG-NAT deployments gets more expensive than implementing IPv6...
It's funny, I would have expected the countries who got the lion's share of IPv4 addresses to lag behind in IPv6 adoption, but it's the other way around.