Metric Anomaly Detection

About Anomalies

Moesif will automatically monitor all your API end points to detect anomalies using both your current and historical data.

Metrics Summary

For each end point (route and verb combinations), Moesif will monitor metrics such as:

  • count of errors, count(distinct(errors))
  • average elapsed time of each invocation, avg(duration_ms)
  • count of events, count(*)
  • count of distinct session tokens, count(distinct(session_token))
  • count of distinct sessions that had an error, count(distinct(errors.session_token))

The summary page should the latest stats for each for the metric metrics, plus any anomalies detected within last 12 hours.

Metric Anomaly Detection (Time Series)

For each of the metrics tracked, you can generate a time series chart with the anomalies highlighted.

Anomaly Detection Time Series

Notes:

  • The maximum date window is limited to 3 days except for Enterprise level accounts.
  • Each of the tracked metric is aggregated at 5 minute intervals. If during a particular 5 minute interval, there are no APIs call for that particular end point, then that time slot will not appear as a datapoint on the time series chart.
  • You can also set alert rules for receiving notifications.

Updated: