Organize a Fabric Lakehouse using Medallion Architecture Design
Organize a Fabric Lakehouse using Medallion Architecture Design
Source
- Provider: Microsoft
- Platform: Microsoft Learn
- Source title: Organize a Fabric lakehouse using medallion architecture design
- Source URL: https://learn.microsoft.com/en-us/training/modules/describe-medallion-architecture/
- Verification Link: https://learn.microsoft.com/api/achievements/share/en-us/pcoletsos/BUF2MU4D?sharingId=F7F61A3006E08628
- Content type: Self-Paced Training Module & Achievement Badge
- Target Exam: Exam DP-700
- Date completed: 2026-08-04
- Last reviewed: 2026-08-15
Summary
This module establishes the enterprise design principles of the Medallion Architecture (Bronze $\rightarrow$ Silver $\rightarrow$ Gold) within Microsoft Fabric Lakehouses to incrementally clean, curate, validate, and enrich data assets for high-performance consumption.
Medallion Architecture Breakdown
[External Sources] │ ▼ (Append-only / Raw ingestion)┌────────────────────────────────────────────────────────┐│ 🥉 BRONZE (Raw Zone) ││ - Raw, unvalidated, historical audit trail ││ - Formats: Parquet / Delta Lake ││ - Ingestion: Data Factory Pipelines, Eventstreams │└──────────────────────────┬─────────────────────────────┘ │ ▼ (Cleaning, Deduplication, Schema Enforcement)┌────────────────────────────────────────────────────────┐│ 🥈 SILVER (Cleansed / Conformed Zone) ││ - Standardized schema, normalized relationships ││ - Deduplicated, cleansed strings, UTC timestamps ││ - Engine: PySpark Notebooks / Dataflows Gen2 │└──────────────────────────┬─────────────────────────────┘ │ ▼ (Star Schema, Aggregations, Business Logic)┌────────────────────────────────────────────────────────┐│ 🥇 GOLD (Curated / Business Zone) ││ - Star schema: Fact & Dimension tables ││ - Optimized for Direct Lake reporting & Analytics ││ - Engine: Fabric Data Warehouse / Lakehouse SQL │└────────────────────────────────────────────────────────┘Layer Characteristics in Microsoft Fabric
| Layer | Primary Role | Data Structure | Typical Ingestion Engine | Target Consumers |
|---|---|---|---|---|
| Bronze | Raw data preservation | Append-only raw Delta tables or files | Data Factory Copy Activity, Eventstreams | Data Engineers, Reprocessing Pipelines |
| Silver | Cleaned enterprise view | Conformed, deduplicated Delta tables | PySpark Notebooks, Dataflows Gen2 | Data Engineers, Data Scientists |
| Gold | Business aggregations & BI | Star Schema (Dimensions & Facts) | PySpark Notebooks, Fabric Data Warehouse | Power BI, Business Analysts, Executive Dashboards |
Multi-Workspace vs Single-Lakehouse Topology
- Single Lakehouse with Layer Schemas: Low complexity, suitable for small to mid-sized analytical projects.
- Multi-Lakehouse (Bronze LH, Silver LH, Gold LH): Enhanced governance, distinct workspace security boundaries, independent compute/storage scaling.
Exam Traps & Gotchas
[!IMPORTANT]
- Bronze is Append-Only: Never perform destructive schema mutations or updates in the Bronze layer; it serves as the ultimate historical recovery source.
- Direct Lake Compatibility: Ensure Gold layer tables are fully structured Delta tables with valid relationships and without complex calculated DAX columns on the model side, enabling full Direct Lake acceleration.