Lakebase Compute and Storage Costs
Open Lakebase in the sidebar to investigate managed PostgreSQL spend across projects, branches, and compute endpoints. Start by selecting a date range and workspace filters.
This is an analytics view of Lakebase resources in your Databricks environment, separate from the database used to host LakeSentry.
Data requirements and cost model
Section titled “Data requirements and cost model”Lakebase analytics depend on ingested billing usage and pricing with Lakebase resource identifiers. Complete connector setup and confirm billing extraction before investigating an empty page.
Compute and storage are separate components: compute usage is measured in DBUs and storage in DSUs. LakeSentry treats both as Databricks cost. Cloud Only mode therefore shows zero for these charges. Use Databricks Only, Total, or Breakdown to view them.
| Tab | Use it to |
|---|---|
| Overview | Compare total, compute, and storage cost; inspect trends and project-level spend |
| Compute | Compare endpoints, billed activity, and compute costs |
| Storage | Inspect branches and storage composition by billing type |
| Optimization | Review open insights, always-on endpoints, branches with no or low compute, and storage history costs |
Costs follow the selected range. Runtime status is classified from billed usage over a fixed trailing 30-day window, independently of the selected range.
Endpoint statuses include Always-on, Intermittent, Low use, Inactive, and Observing. Branch statuses are Active, Low compute, and No compute. These describe different things: an endpoint’s runtime pattern versus compute activity associated with a storage branch. They are not live database availability or CPU measurements.
Investigating an opportunity
Section titled “Investigating an opportunity”- In Overview, identify the projects contributing most to spend.
- In Compute, inspect costly endpoints and their billed activity.
- In Storage, check branches that continue to incur storage cost with little or no compute.
- In Optimization, review the evidence and open the related insights. The Insight Type Catalog describes each Lakebase finding.
- Confirm workload requirements and resource state in Databricks before resizing, suspending, or deleting anything.
Storage composition can distinguish data from history. Where history or data cost is shown separately, it is apportioned from branch cost using the DSU share, rather than an independently metered price for that component. A branch with no recent compute may still hold required data or recovery history.
Troubleshooting
Section titled “Troubleshooting”| Symptom | Check and next action |
|---|---|
| No projects, endpoints, or branches | Widen the date range, clear filters, and verify Lakebase billing usage was extracted |
| Costs disappear in Cloud Only mode | Switch to Databricks Only or Total; Lakebase charges are Databricks cost |
| Status differs from activity in an older range | Status uses the trailing 30 days, while cost uses the selected range |
| Storage cost remains after compute stops | Storage and compute are billed separately; inspect the branch and its retained history |
| Analytics unavailable | Retry, inspect Data Freshness, and include the affected project or branch in a support report |