Kimball Dimensional vs. Data Vault 2.0 in the Medallion Gold Tier
Navigating Star Schemas, Auditability & One Big Table (OBT) in Fabric & Databricks
Data Vault 2.0 is an ingestion and enterprise audit engine best deployed in the Raw/Silver tier; Kimball Dimensional modeling remains the premier consumption layer in the Gold tier for DirectLake Power BI and self-service analytics.
Executive Highlights:
- Enterprises attempting to expose raw Data Vault Hubs, Links, and Satellites directly to BI tools face massive join penalties and severe business comprehension friction.
- Kimball Conformed Dimensions and Star Schemas remain unmatched for VertiPaq compression, Power BI relationship modeling, and intuitive self-service querying.
- One Big Table (OBT) patterns work well for specialized lakehouse ad-hoc exploration, but cause severe data duplication and slow down tabular relationship navigation when scaled across dozens of business domains.
Explore this brief's interactive React Flow topology & weighted trade-off matrix
Architectural Drivers & Industry Dynamics
Regulatory Auditability vs. Serving Speed
Banking and sovereign government agencies require append-only historical audit trails (Data Vault), whereas business users demand sub-second dashboard aggregations (Kimball).
DirectLake Zero-Copy Constraints
Microsoft Fabric DirectLake operates fastest on clean Star Schemas where 1-to-many relationships map directly into memory without complex SQL views.
Schema Agility & New Source Onboarding
Data Vault allows new source systems to be added in parallel without modifying existing tables; Kimball requires careful conformed dimension refactoring.
Side-by-Side Trade-off Matrix
| Architectural Metric | Kimball Star Schema | Data Vault 2.0 | One Big Table (OBT) |
|---|---|---|---|
| Primary Medallion Tier | Gold (Consumption & Serving) | Silver / Cleaned (Enterprise Integration) | Gold (Ad-hoc exploratory marts) |
| Business Comprehension | Intuitive (Facts & Dimensions) | Complex (Requires mart projection layer) | Moderate (Wide, repetitive columns) |
| Auditability / Cryptography | SCD Type 2 business history | Append-only hash keys & load timestamps | Weak historical trace |
| DirectLake Power BI Fit | Optimal (Fastest VertiPaq compression) | Poor (Multi-table joins require bridge views) | Good (Zero joins, but large memory footprint) |
| Parallel Team Scalability | Moderate (Conformed dimension coordination) | High (Independent hub/link development) | Low (Single table lock-in) |
1. The Direct Query Trap: Exposing Data Vault to BI
2. The Modern Harmonized Architecture
3. DirectLake and V-Order Synergy with Kimball
The Architectural Verdict
Deploy Data Vault 2.0 in the Silver tier to insulate your data platform from source system churn and satisfy regulatory audit mandates. Project into Kimball Star Schemas in the Gold tier to deliver lightning-fast, intuitive Power BI DirectLake reporting.
Recommended Action Items:
- ✓Audit your current Gold consumption layer: identify if complex multi-join views are choking Power BI dashboard responsiveness.
- ✓Standardize transformation logic using dbt Core to automate the projection from Silver Data Vault into Gold Kimball marts.
- ✓Benchmark Parquet memory footprints between Kimball star models and wide denormalized tables in OneLake.
Ready to implement or review this architecture?
Available for Principal Contract placements, milestone SOWs, and fixed-scope architecture audits in Sydney & Canberra.