Skip to main content

Compute Costs

The Compute Costs page at /finops/compute attributes DBU consumption and cluster spend to specific compute resources: SQL warehouses, all-purpose clusters, job clusters, DLT pipelines, and serverless endpoints.

SKUs covered

SKUDBU rate (relative)Use case
SQL ProMediumWarehouses (standard)
SQL ServerlessHighWarehouses (serverless)
All-PurposeHighNotebooks, dev clusters
Jobs ComputeLowScheduled jobs
DLT Core / Pro / AdvancedVariesDLT pipelines
Serverless JobsMediumServerless job runs

Cards

  • Total compute DBUs
  • Total compute cost
  • Most expensive SKU
  • DBU/hour effective rate

Breakdown

The page splits costs across:

  • By SKU — pie chart + table
  • By resource — top warehouses, clusters, pipelines by cost
  • By owner — top users / teams burning DBUs
  • By time-of-day — bar chart of DBUs/hour to spot off-hours waste

Drilldown

Click any resource to open a panel with:

  • Daily cost trend
  • Avg / peak utilisation
  • Auto-stop / auto-terminate setting
  • Right-sizing recommendation (if applicable — see Recommendations)

Right-sizing

The page flags resources where avg utilisation is well below capacity:

  • All-purpose clusters with avg < 30% utilisation → suggest smaller node type
  • Warehouses running 24/7 with idle gaps → suggest auto-stop
  • DLT pipelines with high task slot waste → suggest fewer max workers

Apply via the Apply recommendation button (writes the change back to the resource via Databricks API).

API

GET /finops/compute?days=30
GET /finops/compute/{resource_type}/{resource_id}
POST /finops/compute/right-size { "resource_id": "...", "target_node_type": "..." }