IPv6 vs. IPv4 Distribution According to My Personal Browsing History

I read an article about the still too slow process of people supporting IPv6 on their servers. Because of the ever increasing IPv4 shortage, Amazon Web Services recently startet to charge a fee for each IPv4 address their customers use.

Because there are different distributions of IPv6 vs. IPv4 floating around, I wanted to know how my personal experience looked like.

I exported my personal web history from my web browser for the last year and saved all 1,800 unique domain names in a `history.txt` file. As I currently work from home, almost all my webpage access happens in the same browser. Therefore my personal history should be quite complete, minus the obviously missing domains from surfing in private mode.

I created a small shell script to find out if the domain names are served over IPv6:

#!/usr/bin/env sh

for url in $(cat history.txt); do
  aaaa=""
  aaaa="$(dig +short aaaa ${url} | grep ':')"
  if [ "${aaaa}" != "" ]; then
    echo ipv6 >> history.log
  else
    echo ipv4 >> history.log
  fi
done

The result is 50,67 % of my visited domains are served over IPv6 and maybe IPv4, while 49.33 % are only served over IPv4.

Looking at what kind of domains are not served over IPv6, there were some surprises. Even big technology companies like GitHub don’t serve their most important domains over IPv6, which is very disappointing to say the least.

Some companies are IPv4-rich for historic reasons, which might me acceptable. But running services without IPv6 support nowadays is gross negligence in my opinion.

In Geminispace, the distribution is much worse. I didn’t measure it myself, because we already have a crawler checking the whole space. It’s a sobering 18 %. People hosting from home might account for the difference.

Lupa’s Statistics (Search for IPv6)

-- Response ended

-- Page fetched on Mon May 6 21:49:42 2024