Google IPv6 stats (approaching 50%, ever so slowly): https://www.google.com/intl/en/ipv6/statistics.html
V6 addresses are not friendly. Four numbers is east to communicate, easy to enter, and easy to spot even for a non technical person. V6 is more difficult and variable length which is harder to explain to your grandparents. I also think there is something to be said about people with dyslexia and how the numbers are perceived by them.
And there will be lots of support needed to roll this out to home routers with various levels of support and bugs, needing to enable firewalls at the device level instead of at the router, all of the IoT devices with just enough processing power to send an up address on a half baked networking stack. It’s still a massive uphill battle. And yes, I know there are solutions to all of these issues, it’s just getting them out there.
URL's don't have a fixed length, so I'm not sure variable length is significant barrier to adoption by itself? Phone numbers also have variable lengths.
Yes in many parts of the world you are correct. In North America, phone numbers are all 10 digits prefixed with a 1 for long distance.
> URL's don't have a fixed length, so I'm not sure variable length is significant barrier to adoption by itself?
URLs tend to be a word or phrase that can be communicated. You tend to have to just say google dot com not gee oh oh gee el ee dot cee oh em.
911, and other special and vanity numbers don't count? People have never heard of international numbers, either? Or that long distance vs local changes the length?
Also basically every home router in the last 10 years is V6 dual stack because they're all more or less just running variations on Linux or FreeBSD with user land stripped out for Busybox.
Ipv6 has been fine for years. It is purely institutional laziness at this point.
I more or less agree that it should be hands off. However explaining to someone over a voice channel how to enter an IPv6 address who isn't familiar even once is unfriendly. Not impossible, more difficult than IPv4.
> Also basically every home router in the last 10 years is V6 dual stack because they're all more or less just running variations on Linux or FreeBSD with user land stripped out for Busybox.
Yes, however they all have various levels of support for managing IPv6. Many early routers didn't have friendly interfaces to set this up, or did so automatically if the ISP provided addresses. Firewall support is still a mixed bag IMO, especially when you start to consider the crap router/gateways that ISPs provide to the lowest bidder. Good hardware with good support isn't something a non-technical person with a single tablet in the house is going to pick up next to the 40-Dollar option on a shelf. You are right, it's getting better.
> Ipv6 has been fine for years. It is purely institutional laziness at this point.
You could also argue that despite the dire warnings about the impending doom of the internet, IPv4 has largely been 'fine' for years as well, and will likely continue to hamper IPv6 adoption until v4 is no longer 'fine'.
Also who are you explaining entering v6 addresses to? Serious question. And how is that effectively longer than a dns name?
People aren't used to v6 addresses, I get that. But they aren't meaningfully harder than myts3server.cgnatsucksdns.com.
Your example of 8.8.8.8 is itself designed to be easy to communicate and enter, same with 1.1.1.1 and 9.9.9.9, etc. They exist because communicating an ISPs DNS server or a standard IPv4 address can be hard sometimes.
IPv6 addresses are still fixed size. They aren't variable length. The :: shortcut is just "fill with zeroes here" and your grandparents probably had lots of bank account numbers and credit card numbers they memorized as prefix, suffix and some number of zeroes in the middle. (Like the :: shortcut they might not even bother to remember how many zeroes and just fill out the boxes until it looked right.) That's still common among some banks to use account/card numbering schemes like that, even in the computer age where you probably shouldn't memorize your account number and it is better for security if the account number had something a lot more pseudo-random than a long series of zeroes in the middle.
> I also think there is something to be said about people with dyslexia and how the numbers are perceived by them.
I can't speak for anyone else with dyscalculia (like dyslexia but only for the ordering of numbers) but hexadecimal is a lot kinder to me than decimal. I often remember the right digits but not their order, but I rarely have such problems with "words" and IPv6's use of hex 4-digit groups looks to my brain a lot more like words than numbers, especially when a through f show up. I hate trying to remember IPv4 addresses in dotted quad format because it is sometimes a million ways to get the orders wrong even knowing the format, but I've personally had fewer problems visually/memorably with IPv6 addresses, especially when you get a nice chunky :: in there somewhere. (It's so great sometimes to just handwave away all the less meaningful zeroes.)
Again, that's just my experience with my own (undiagnosed from a professional standpoint, undebilitating) dyscalculia.
Edit: I just remembered seeing something similar related to rounding a floating point signal once. Maybe that is the reason behind it.
Edit 2: Peaks seem to fall on Saturdays, so maybe this corresponds to people turning on their computers/routers at home more often on the weekend.
(At least as I've heard it explained before. But it could be a large country with high adoption like India swaying the statistics.)
It's reflective of work-in-office patterns--IPv6 penetration is much lower in business settings than in residential (or mobile) settings.
I'm surprised that it falls back pretty equally on Fridays and Sundays, though. Maybe this is an artefact of the timezone they are operating in - some of the Friday traffic is actually Saturday, some of the Sunday traffic has gone to Saturday.
Other comments are pointing out it's much much easier to migrate mobile to ipv6 than it is to migrate desktop, so I'm thinking it's more likely this than the other possibilities.
Worldwide. For the US specifically it's 52%:
* https://www.google.com/intl/en/ipv6/statistics.html#tab=per-...
- FTTH everywhere (even in remote areas) with provisions to ensure competition and eliminate ISP lock in (which happened in the cable days, lessons were learned)
- tracking IPv6 deployment, publishing detailed yearly reports of status and plans, and pressuring ISPs
- mandating IPv6 if a mobile operator wanted a 5G license
Net result is you can basically assume that IPv6 is there except in increasingly rare cases (90-99% except one laggard). IPv6 just works, and is a freaking breath of fresh air in a world of double NAT and CGNAT. I've been on IPv6 for over a decade with both simple and complex scenarios, and every bit of IPv6 is just flat out better than IPv4 in my experience.
I wrote a lengthy comment a month ago regarding the ARCEP IPv6 report: https://news.ycombinator.com/item?id=36788130
Kidding, but also kinda not. We should make ipv7 which is just ipv6 but the extra bytes go in an expanded header, like the author suggests.