the ipv4 vs ipv6 growth debate

I use a web plugin tool called ipvfoo to track my actual ipv4 vis ipv6
usage. I wish it worked over time. With very few exceptions I am still
regularly calling ipv4 addresses in most webpages. Has anyone done a
more organized study of say, the top 1 million, and how many still
require at least some ipv4 to exist, and those trends over time?

I use the same extension on Chrome.

I’m surprised that with all the recent hoopla about it, from the major social media platforms, Twitter still shows serving their http site over IPv4, Facebook and LinkedIn show solid IPv6.


It would be nice if IPvFoo showed the bytes and connection/request count. It's going to be a loonnggg time before we can do consumer internet browsing with no v4, until then it's about reducing cost of CGNAT with reduced packets/connections.

For twitter, the main site is v4, yea, but (Edgecast) and (Fastly) make up the vast majority of the bytes fetched on the site for me and are both v6 now. I don't recall when I last checked but they were still v4-only not too long ago.

The other end of it is v6-only servers that don't accept inbound connections. Thos have been hampered IME by github not serving git over v6. Supposedly it's coming soon but so much modern software fetches stuff from Github that that's a major blocker.


Often lost in the ‘debate’ about V6 adoption is that for a 100% native IPv6 experience to work, there are multiple other components that have nothing to do with the network that ALSO have to work correctly. Any issues with these are likely going to cause fallback to v4.

It’s very difficult to know how much v4 traffic to a website COULD have worked just fine on v6, but didn’t, and why it didn’t.

With IPv6Foo you can click on the icon and it will show you a table listing what URLs are serving some piece of a given page with v6 and v4.

LinkedIn for example shows the main feed page served via v6 but there are a couple of pieces with v4 from these sites

Some may be feeding ads content, others tracking, market research, etc.


But IPv6Foo , ast least as far as I could tell by quickly looking at the code, cannot tell you if an IPv6 connection WOULD have worked, but IPv4 is where it ended up.

With Happy Eyeballs, if the IPv4 TCP session finishes up only a couple ms faster than the IPv6 ones, the v4 one wins out. That doesn’t give you any meaningful signal as to WHY it landed on IPv4 instead.

Well hard for them to establish an ipv6 connection, none of the domains for the urls I posted have an aaaa record :slight_smile:


Of course for those, yes. I was speaking more generally. Sorry if that wasn’t clear.

If you really want to know if a site works over IPv6 you have to flush any cached pages and turn off IPv4. You also need to be using a IPv6 only nameserver. Only after doing those extra steps can you say a site is IPv6 ready. I’ve had pages that appeared to be all IPv6 fail after doing these extra steps.

As for connection racing IPv6 wins 99.99% of the time. There is enough bias that it will win unless there is a lossy path involved.