Anomaly monitors

We’re happy to announce the introduction of anomaly monitors.

Anomaly monitors allow you to receive an alert when the results of your query deviate from what’s expected. In comparison to threshold monitors, which compare the results of your query against a fixed threshold, anomaly monitors use the recent behavior of the query to determine if the most recent results are anomalous.

For example, suppose you want to monitor response times for a range of API endpoints that you serve. Rather than identifying the correct threshold for each one, you can simply use an anomaly monitor to be alerted whenever there’s a sudden spike in response time. Anomaly monitors allow you to group the results of your query and track expected behavior for each group separately.

For more information, see Anomaly monitors in the documentation.