Backstage with Lakebase, part 3
Databricks demonstrates how Lakebase and Unity Catalog enable a single SQL query to join operational ownership data with cloud billing, eliminating ETL pipelines and improving cost visibility for ephemeral database branches.
The useful question is what changes for users, developers or buyers, and whether the announcement stays industry context or becomes something people can actually use.
In the third part of the series, Databricks shows how Lakebase’s isolated compute per workload allows a FinOps analyst to query Backstage’s operational data alongside billing data without impacting live portal performance. The demonstration uses Lakehouse Federation to join Backstage’s Postgres data with Unity Catalog’s billing rows in system.billing.usage, producing a single query that historically required separate systems and data movement. The approach leverages Lakebase’s one-second database branching and Unity Catalog’s governance to provide real-time cost attribution for temporary environments.
The practical impact of this integration is immediate visibility into infrastructure ownership and cost, replacing manual processes like ETL pipelines or Jira tickets. The workaround for Lakehouse Federation’s Postgres connector involves creating a native Postgres role with static credentials, managed separately from the OAuth identity used by applications. While this requires password rotation and additional setup, Databricks notes it aligns with security best practices by isolating federation workloads from application users.
Beyond ownership queries, the billing data reveals development process costs by tracking compute usage per branch and endpoint. Ephemeral branches created for pull requests or feature development appear as line items in system.billing.usage, enabling engineering managers to assess the cost of branching workflows against productivity gains. The test branch in the proof of concept incurred a cost of 0.0107 DBU, illustrating how minimal overhead can be attributed to temporary environments that scale to zero when inactive.
The series concludes with a call to action for teams to adopt this pattern, emphasizing the measurable benefits of integrating operational and analytical data. Databricks and Thoughtworks are collaborating to transition this proof of concept into production use, with plans to measure real sprint velocity and development outcomes. The convergence of operational and analytical systems, once separated by decades of infrastructure silos, now enables teams to define new standards for cost transparency and development efficiency.