Observability

Lakebase gives you two complementary ways to observe your Postgres databases:

• Advanced Postgres telemetry — Let AI find and fix problems for you. Capture deep Postgres signals like query plans, session history, and schema changes as Delta tables in Unity Catalog, where Insights flags problems on a schedule and Genie helps you investigate and fix them. Prebuilt dashboards visualize the same data.
• Real-time monitoring — Built-in views of current Postgres health and activity: metrics, running queries, recent query performance, and system operations. Available on every project by default.

To turn on advanced telemetry, configure observability once and assign it to your projects. Real-time monitoring is always on and needs no setup.

Advanced Postgres telemetry

Advanced telemetry records deep Postgres signals like query plans, session history, and schema changes, writing them to Delta tables in your own Unity Catalog. This captured history is what lets you troubleshoot with AI: Insights reviews it on a schedule and surfaces issues without being asked, and Genie investigates problems you describe and proposes a fix you confirm. Prebuilt dashboards visualize the same data, and because it lands as Delta, other Azure Databricks tools can read it and it persists across restarts. Configure observability to choose the destination and the identity that writes it.

Note

Use this five-minute quickstart to set up observability on a single project with your own credentials, so you can see how it works before you roll it out more broadly.

Tool	Description
Configure observability	Set the telemetry destination, choose the identity that writes it, and roll it out across your projects.
Insights	Reviews your telemetry on a schedule and surfaces issues without being asked, so you don't have to keep checking.
Genie	Investigates a problem you describe conversationally and proposes a fix you confirm.
Telemetry dashboards	Prebuilt dashboards over your captured telemetry, including query detail and time-interval comparison.
What gets captured	Reference for every Delta table and column Lakebase writes to Unity Catalog.

Real-time monitoring

Open a project's Monitoring page in the Lakebase App to see what your database is doing, scoped to the branch and compute you select. The built-in views read your compute's metrics and Postgres's own statistics directly: metrics graphs go back up to seven days, while the query views are live and reset when a compute suspends.

This same page also hosts the Advanced Postgres Telemetry tab, where prebuilt dashboards appear once you configure advanced telemetry.

Tool	Description
Metrics	Graphs of a compute's RAM, CPU, connections, database size, cache, and more, over a selected time period.
Active queries	The queries running right now, from pg_stat_activity. Sort by duration to find long-running operations.
Query performance	The top 100 recent queries by calls and execution time, from pg_stat_statements, grouped by normalized statement.
System operations	A log of platform actions on the project: starting and suspending compute, creating branches, applying config.