Metric Anomaly Detection

Metric anomaly detection uses dynamic baselining to automatically detect changes in metrics. This feature runs continually on our systems to detect significant changes in the most important queries. Anomaly detection uses the standard Events feature for reporting, which also allows you to configure alerts using email or any of our standard integrations (Slack, HipChat, PagerDuty, VictorOps, and more) to get notified. The category used for metric anomaly detection is “Anomaly in Metric”.

At the moment, anomaly detection focuses on changes in the following metrics:

  • Frequency and total accumulated time per category of query
  • Overall query error and warning rates, globally (not per-category-of-query)

Anomaly detection is automatically enabled; these metrics apply to all database technologies supported by VividCortex.

Anomaly detection events in the Events Dashboard contain deep-links into the Metrics detail page and the Query details page, allowing you to quickly drill-down and get more information.

Anomaly in Metric event details

Baselining

In order to minimize false-positive alerts, VividCortex uses intelligent baselining to account for seasonalities in metrics. We consider at least hourly, daily, and weekly seasonal trends to avoid alerting on typical cronjobs and other periodic activity.

Due to the nature of baselining, we require at least one week of prior data for each metric for anomaly detection.