Probably two pretty bad days I can recall. One started early morning on a Sunday due to an unannounced GKE upgrade that lasted until 3am that night (thanks Goog

Probably two pretty bad days I can recall. One started early morning on a Sunday due to an unannounced GKE upgrade that lasted until 3am that night (thanks Google). Another was actually due to the polar vortex falling over Houston and knocking out power to some of the primary and secondary cloud provider servers we use to host Kafka. If you’re talking about minor fires (phone calls for network blips or something similarly minor), then yes 20 seems about right.