On December 6th, Monday at 18:39 UTC a maintenance process caused platform instability. At 19:32 UTC we were alerted to zonal degradation and invoked our zone fail-over procedure. At 04:00 UTC all affected sites recovered. Approximately 350 sites were affected for more than 15 minutes, approximately 30 sites were affected for more than 41 minutes and 4 sites were affected for up to 4 hours. As a result of this incident we’ve identified, and are implementing, additional mitigations to help better prevent and recover from incidents like this in the future.