The Women's Football World Cup Final in HTTP access log lines processed (for a subset of our logs).
Funny how literally every football match follows this exact pattern - unless they go to extra time/penalties. #WebStats#WWC#Football
I just compared "all clients" with "likely human users" for percentage of requests to www.bbc.co.uk & www.bbc.com which do not include a TLS SNI header and quite often, the "likely human" percentage is higher. I guess it's easier to keep scripts and bots up to date than web browsers?
Raw data, if anyone has a use for it (totally anonymous, so safe to share): https://gist.github.com/neilstuartcraig/4662c4be91a7afb1b5d479755cf84ddb
Still bends the mind a bit that we serve nearly 14M web pages every day to Search Engine indexers & bots (and those are just the bigger orgs indexers/bots). That's 4% of our traffic.
Also super strange how skewed towards Google this is, they're making 75% (over 10M) of those requests.
Might be interesting to plot the decline of Twitter actually...
Then there's "monitoring" which is everyone's "is the internet working" - nearly 32M of those every day.
Yesterday I added a graph to track TLS ciphersuite usage over time & immediately spotted an anomaly.
In mid-Feb, on our commercial CDN, CHACHA usage dropped from ~10-12% to ~0.5% & stayed there. The same did not happen on our own CDN so it seemed unlikely to be client behaviour.
Raised it with the CDN vendor & they tied it to a release which accidentally changed the behaviour to pref CHACHA for clients for whom CHACHA is top pref.
Having good data is 💯 #TLS#CDN#Grafana#webStats