Skip to content

Databricks Cost Explorer

The Cost Explorer is LakeSentry’s primary tool for ad-hoc cost investigation. While the Overview Dashboard gives you a summary, the Cost Explorer lets you slice costs across multiple dimensions and drill down to the specific resources driving spend.

Period-over-period comparison and trend analysis now live on the separate Cost Changes page. Legacy Cost Explorer URLs with tab=comparison or tab=trends redirect there, and tab=attribution redirects to Cost Allocation.

The Cost Explorer is organized into tabs, each offering a different view of the same underlying cost data.

The default view. Shows cost as a hierarchical bar chart grouped by a dimension you choose, with a detail panel below showing the selected group’s cost breakdown.

Available grouping presets:

PresetHierarchy
By WorkspaceWorkspace → Work Unit
By TypeType → Work Unit
By OrganizationOrg Unit → Department → Team → Principal
By ProjectProject → Team → Principal

Switch presets using the dropdown above the chart. Click any bar to select it and see its detail panel, including cost breakdown and child items.

A dedicated breakdown showing Compute vs. Service cost categories with daily trends. Includes sub-breakdowns by compute type (Jobs, All Purpose, SQL, etc.), service type (DLT, Model Serving, etc.), usage type (Compute Time, Storage, Network), and cloud instance type.

Cost grouped by the users and service principals who ran workloads. Each row shows:

  • Principal name, type (user, service principal, or group), and workspace count
  • Total spend for the period with cost trend
  • Query count and request count
  • Client application breakdown
  • Anomaly count (audit events)
  • Last active timestamp

This view helps answer “who is spending the most?” and is useful for chargeback discussions. Summary charts show the top 10 spenders and aggregate spend trend over time.

Cost attributed to Delta tables through the work units (jobs, pipelines) that write to them. Each row shows the table name, total cost, number of associated work units, last access time, and a staleness status indicator. Summary stats highlight stale table cost and unused table count.

A pivot-table style view that crosses two dimensions, with two available modes:

  • Workloads — Workspace (rows) vs. Workload Type (columns: Jobs, Pipelines, SQL Warehouses, Model Serving, Other). Click any cell to see details.
  • Principals — Principal (rows) vs. Workspace (columns). Click a row to navigate to that principal’s detail page.

Each cell shows the cost, and cells are heat-mapped by magnitude. The Matrix tab is useful for answering questions like “which workspaces have the highest job costs?” or “which principals are active across multiple workspaces?”

All Cost Explorer tabs share a common filter bar:

FilterWhat it controls
Time rangeThe analysis period (presets or custom date range), managed by the global time range selector
WorkspaceLimit to specific Databricks workspaces (dropdown with multi-select)
TypeLimit to specific workload types: Jobs, Pipelines, SQL Warehouses, Serving Endpoints (dropdown with multi-select)

Additionally, deep links from other pages (e.g., budgets) can pre-set organizational filters — Team, Department, Org Unit, and date range overrides — which appear as removable badge chips in the filter bar.

Filters can be combined. For example, filter to “SQL Warehouses” in a specific workspace to see warehouse costs across that workspace.

Active filters are displayed as chips in the filter bar. Click the × on any chip to remove it.

The Breakdown tab’s hierarchical bar chart supports drill-down navigation. Click any bar to zoom into that group and see its children. The drill-down levels depend on the grouping preset — for example, with “By Workspace,” you drill from workspace to individual work units; with “By Organization,” you drill from org unit to department to team to principal.

Each drill-down step adds a breadcrumb to the navigation bar, so you can step back to any level. Clicking a bar also opens a detail panel showing the selected item’s cost breakdown and children.

The Cost Explorer supports URL-based deep linking. Filters, grouping presets, date ranges, and the active tab are encoded in URL parameters, so you can share a specific view with colleagues or bookmark it for later. Other pages (such as Budgets) link directly into the Cost Explorer with pre-applied filters and grouping.

Legacy deep links are preserved: tab=comparison and tab=trends redirect to Cost Changes, while tab=attribution redirects to Cost Allocation.

Like the Overview Dashboard, the Cost Explorer supports the cost mode toggle:

  • Total Cost — Combined Databricks + cloud infrastructure cost
  • Databricks Cost — Databricks costs only
  • Cloud Cost — Cloud infrastructure cost only
  • Cost Breakdown — Side-by-side Databricks and cloud split

The cost mode selection persists across tabs within the Cost Explorer.