Users at a few Canvas institutions experienced slowness in the application between 11:37 AM and 11:50 AM MT this morning. Our DevOps team scaled up key resources in response to an unforecasted surge of user activity and performance returned to normal levels at 11:50 AM.
Canvas is built to auto-scale key resources in response to (1) patterns in historical usage data and (2) real-time usage data. Scaling takes a little time, so the process works more smoothly when current usage patterns roughly match historical data and we can scale preemptively. Our scaling algorithms learn something from every event like today's. We apologize for the slowness that users at the impacted schools saw for about fifteen minutes this morning.