Your disaster recovery plan isn’t going to work! At least not the way you expect. Recent months showed how “perfect” preparations can collapse when a service provider incident hits and reality drifts off the playbook. Sometimes the plan is robust, but the switch that should activate it can’t be reached (did someone say AWS console login page?).

This talk is about the most underrated DR capability: improvisation. I’ll share hands-on, real-world stories of responding to third-party outages in real time - moments where only deep, end-to-end system understanding, plus a steady head under pressure, made recovery possible. You’ll hear about the night I wired my personal laptop into production to stop revenue from leaking, and the time a temporary client-side SSL change helped bypass a CDN outage, plus a DNS cache making everything worse.

You’ll leave with practical tactics for the next outage, and the mindset that chaos doesn’t read your documentation.