FinOps Engineering: System Tables and Chargeback
The Cost Iceberg section taught you where compute spend hides; this section teaches you how to turn
that spend into a number a VP of Finance will sign off on. In a legacy warehouse, chargeback meant a
DBA exporting a spreadsheet of CPU-hours against a cost-center code once a quarter. On Databricks,
the raw ledger is a set of system tables – system.billing.usage, system.billing.list_prices,
and their neighbors – that you can query like any other Delta table, join against your own tagging
taxonomy, and refresh daily instead of quarterly. That shift, from a monthly finance ritual to a
live, queryable feed, is what “FinOps engineering” means in practice: treating cost attribution as a
pipeline you build and own, not a report someone else hands you.
This section walks the full arc: understanding the billing system tables and how DBUs become dollars, designing an attribution taxonomy that survives contact with shared clusters and multi-team warehouses, building a self-serve chargeback dashboard so the conversation moves from “why is our bill so high” to “here’s exactly which team and job drove it,” proving that optimization work saves real money rather than just reclaiming disk, and previewing where automated cost governance is headed next.
flowchart LR
A["system.billing.usage<br/>raw DBU records + custom_tags"] --> B["system.billing.list_prices<br/>time-versioned $/DBU by SKU"]
B --> C["Attribution join<br/>tag → team / project / cost center"]
C --> D["Chargeback dashboard<br/>headline, by-team, drill-down"]
C --> E["system.storage.predictive_optimization_operations_history<br/>proves $ saved, not just bytes freed"]
D --> F["Monthly chargeback report<br/>to Finance"]
E --> F
C --> G["Budgets + guardrails<br/>alert thresholds, tag enforcement"]
G --> F
Lectures
| # | Lecture | Est. Read |
|---|---|---|
| 1 | The system.billing.usage Deep Dive | 3 min read |
| 2 | Attribution and Chargeback Taxonomy | 3 min read |
| 3 | Building the Chargeback Dashboard | 3 min read |
| 4 | Predictive Optimization: Proving Savings, Not Just Space | 3 min read |
| 5 | Auto-TTL: The 2026 Frontier | 3 min read |
| 6 | The System-Table Query Pack and Governance Guardrails | 3 min read |
| 7 | Check Your Knowledge | Quiz |
| ← Previous: Check Your Knowledge | Next: The system.billing.usage Deep Dive → | |:—|—:|