Pattern community advanced
Data Vault (hubs, links, satellites)
Integrating many changing sources into a warehouse without constant rework.
At a glance
| What | Integrating many changing sources into a warehouse without constant rework. |
|---|---|
| When | Enterprise integration of many sources; Strong audit/lineage requirements |
| Where | Data Warehouses |
Problem
Integrating many changing sources into a warehouse without constant rework.
Solution
Model the raw integration layer as insert-only: hubs — business keys links — relationships between hubs satellites — descriptive attributes over time Loads are insert-only and parallel, so a source change adds a satellite rather than forcing a rewrite. Build star schemas as a downstream, disposable layer.
When not to use
- A single source feeding a known set of reports
Security scan
Last sweep 2026-10-24: 0 finding(s), 0 marked as examples.
Context
Many source systems, frequent schema changes, auditability requirements.
Forces
- Sources change constantly
- History and auditability are mandatory
- Loads must be parallel and restartable
Tradeoffs
| Choose this when | Costs & caveats |
|---|---|
| resilient to change, auditable, parallel loads | more objects, harder to query directly; needs a downstream marts layer |
Diagram
source ─▶ hub_customer ─┐
hub_order ─┼─ link_order_customer ─▶ sat_* ─▶ star marts
source ─▶ hub_product ──┘
Registry stats
| Verification | community · Sep 14, 2026 |
|---|---|
| Completeness | 75 |
| Trust multiplier | ×1.25 |
| Composite | 34.75 · adequate |
| Freshness | 40 days since last source check |
| Difficulty | advanced |
| Version | 0.9.0 |
| License | CC-BY-4.0 |
| Security flags | none |
Did this solve your problem?
Data Vault (hubs, links, satellites)