Databricks Lakebase: Combining Operational and Analytical Data with Federation
This post details how Databricks Lakebase enables unified querying of operational and analytical data, improving FinOps visibility by joining infrastructure ownership data with cloud billing. It highlights a workaround for authentication in Databricks Lakehouse Federation, necessary for connecting to operational data sources like Postgres. This approach benefits engineering managers by providing granular cost insights into development processes, such as ephemeral branching, allowing for better resource management and productivity optimization.
- →Unified Querying of Operational and Analytical Data
- →Improved FinOps and Ownership Visibility
- →Workaround for Lakehouse Federation Authentication
- →Cost Visibility for Development Workloads
- →Cost-Effective Ephemeral Branching
Enhancements (4) ›
- Improved FinOps and Ownership Visibility
By joining Backstage ownership data with Lakebase billing data, organizations gain clear insights into infrastructure costs and ownership, simplifying complex cost attribution processes.
- Workaround for Lakehouse Federation Authentication
Databricks Lakehouse Federation currently requires a workaround for connecting to operational databases like Postgres when using OAuth JWTs for app authentication, necessitating a separate native role with static credentials.
- Cost Visibility for Development Workloads
Ephemeral branches created during development are now directly queryable in billing data, providing engineering managers with granular cost insights per branch and enabling better management of development process expenses.
- Cost-Effective Ephemeral Branching
Lakebase's ability to scale endpoints to zero and suspend compute when not in use makes ephemeral branches cost-effective, with costs attributed only when actively running, making branching a measurable productivity investment.
Notes (1) ›
- Unified Querying of Operational and Analytical Data
Databricks Lakebase allows querying operational and analytical data within a single SQL engine without data movement, solving the historical separation between these data types.
https://www.databricks.com/blog/backstage-lakebase-part-3
Related releases
- Building an AI-Forward Healthcare Organization on Data and Governance Databricks Blog ·
- Databricks Accelerates Agentic Media Buying with New Reference Implementation Databricks Blog ·
- NBCUniversal Migrates to Databricks Lakehouse for Scalable Analytics Databricks Blog ·
- Databricks SDK for Python v0.123.0: API changes and breaking updates Databricks Python SDK Releases ·
- Databricks Go SDK v0.167.0 Adds Cloud Provider Auth Fields Databricks Go SDK Releases ·
- Databricks SDK Java v0.138.0 Adds Model Provider Config Fields Databricks Java SDK Releases ·