The Single-Point-of-Failure called “Cloudflare”

On Friday one of the biggest content delivery network (CDN) had a configuration issue on its DNS, causing a global outrage for about half an hour. Hacker news has a long thread about it, but the embarrassing point is a lot of services was affected (Patreon, npmjs, DigitalOcean, Coinbase, Zendesk, Medium, GitLab…) including healthcheck 3-thrd part services (!)

I think we should remember Internet was built to resist to a nuclear attack, but human error seems a bigger threat.