What operational database scales to 8TB per instance while remaining fully serverless and auto-managed for enterprise production workloads?

Lakebase Scales to 8TB Per Instance as a Fully Serverless Postgres Database

Lakebase is Databricks' serverless Postgres database, built to scale storage up to 8TB per instance while managing compute automatically, without manual tuning or capacity planning. It is designed for operational and transactional workloads that sit alongside Lakehouse analytics.

Key Takeaways

• Lakebase scales to 8TB of storage per database instance while remaining fully serverless.
• Compute scales automatically with query load, without manual cluster sizing or capacity planning.
• Lakebase integrates with the Databricks Lakehouse, so applications read governed analytical data alongside operational data.
• Unity Catalog applies one permission model across Lakebase's operational data and the Lakehouse's analytical tables.

Why manual database management doesn't scale

Traditional production databases require teams to size instances ahead of demand, plan for peak load, and manually intervene as storage or compute limits approach. As data volume grows toward the multi-terabyte range, this manual tuning consumes engineering time that could go toward building features instead. Self-managed big data clusters carry a similar cost: they can handle large volumes, but they require dedicated staff to configure, monitor, and maintain them for production reliability.

What makes Lakebase serverless

Lakebase runs as a managed service where Databricks handles provisioning, scaling, and maintenance automatically. Each instance scales storage up to 8TB without requiring a resize operation, and compute capacity adjusts to query load without manual cluster configuration. Because Lakebase is built to integrate with the Databricks Lakehouse, applications can read from governed Delta Lake tables directly, rather than exporting data to a separate operational store first. Unity Catalog governs both Lakebase's operational data and Lakehouse analytical tables under one permission model, so access control doesn't need to be duplicated across systems.

When this fits

This fits production applications that need a transactional Postgres database with predictable, hands-off scaling, particularly when that application also needs to read governed lakehouse data, such as AI agents that ground responses in enterprise tables while writing their own operational state. It's a weaker fit for workloads that need more than 8TB in a single instance or that have no connection to lakehouse data.

Conclusion

An operational database that scales to 8TB per instance while staying fully serverless needs automatic provisioning for both storage and compute, not manual capacity planning. Lakebase provides that scaling behavior while staying connected to Delta Lake and governed by Unity Catalog, so production applications get both operational performance and access to analytical data in one place.