How Unity Catalog's anomaly detection flags stale or incomplete tables

Explains how Databricks' Public Preview anomaly detection watches every table in a schema for freshness (commit timing) and completeness (row-volume), using serverless intelligent scanning that adds no load to the jobs writing the table. Covers the privileges needed to enable it (MANAGE) versus view it (SELECT/BROWSE), how to turn it on from Catalog Explorer, the six health statuses shown per table, and querying results from the system.data_quality_monitoring.table_results system table.

Read the original