Generally available: Azure Monitor metric alerts dynamic thresholds outage detection

Azure Monitor metric alerts with dynamic thresholds detection, leverages advanced machine learning (ML) to learn metrics' historical behavior and identify patterns and anomalies that indicate possible service issues.

Automatic outage detection improves thresholds training after service issues are detected.

Previously, after long service issues were detected, thresholds could have been affected for a long period of time. The new capability automatically recognizes prolonged outages and removes them from threshold training. The results are much better thresholds that fit the data and can detect service issues with the same sensitivity as before the outage.

Before the new capability:

After the new capability:

Read the metric alerts documentation to learn more.