Monitoring
Watch Lakehouse health and usage.
Your monitoring scope is the Lakehouse. You watch whether it is healthy, performing, and being used as intended. The shared Observability model places this alongside the Tenant Admin's resource view.
What to watch
Service health. The Lakehouse runs a catalog and query services; confirm they are up and responsive. Failures here surface to every product that reads or writes the Lakehouse.
Metastore health. The PostgreSQL catalog metastore is the Lakehouse's source of truth for table metadata. Watch its availability and growth; a metastore problem looks like tables "disappearing" or scans failing.
Storage growth and table sprawl. Track object-storage growth and the number of tables and snapshots. Iceberg accumulates snapshots; plan maintenance so storage and query planning stay efficient.
Usage. Which namespaces and tables are queried, and by whom, tells you what to optimize and what is abandoned.
Operating practices
Set a maintenance rhythm for Iceberg tables (snapshot expiry, compaction) so query planning and storage stay healthy as data grows.
Watch the metastore database you dedicated to the Lakehouse; it is the single most important dependency.
Tie usage back to structure. Heavily queried tables may deserve their own optimization; unused ones are candidates for retirement after checking with their owners.
When a problem is Instance-wide rather than Lakehouse-specific, escalate to the Tenant Admin or Operator, whose scopes sit above the Lakehouse.
Last updated
Was this helpful?