“This was our error and not the result of an attack or malicious activity.”
A BGP policy change knocked out 19 data centers for over an hour. Those 19 carry a large share of Cloudflare’s global traffic, and they were the ones rebuilt to be more resilient. The resilience project caused the outage. Cloudflare owned it the same day, which is more than most providers manage. A big slice of the web still went dark because of one company’s config push.