Medallion Architecture Cheat Sheet

Updated 2026-04-12

Next Topic: Prefect Data Orchestration Cheat Sheet

The Medallion Architecture is a layered data design pattern that organizes lakehouse data into Bronze (raw ingestion), Silver (cleaned and conformed), and Gold (business-ready aggregates) layers to incrementally improve data quality and structure. Originating from Databricks best practices, this pattern has become the de facto standard for modern data lakehouses across platforms like Microsoft Fabric, Snowflake, and AWS, enabling teams to build auditable, scalable data pipelines with clear separation of concerns. The key mental model: each layer represents a progressive refinement contract—Bronze preserves raw truth, Silver enforces cleanliness and standardization, and Gold optimizes for consumption, making data quality improvements explicit and reversible rather than buried in opaque transformation logic.

What This Cheat Sheet Covers

This topic spans 20 focused tables and 125 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Core Architecture LayersTable 2: Bronze Layer Ingestion PatternsTable 3: Bronze Layer Design PrinciplesTable 4: Silver Layer Transformation StandardsTable 5: Silver Layer Quality EnforcementTable 6: Gold Layer Business AggregationsTable 7: Schema Evolution HandlingTable 8: Performance Optimization TechniquesTable 9: Data Quality Layers AssignmentTable 10: Multi-Hop Architecture PatternsTable 11: Tooling and FrameworksTable 12: Platform-Specific ImplementationsTable 13: Access Control and GovernanceTable 14: Common Anti-Patterns and MistakesTable 15: Optimization Best PracticesTable 16: Naming ConventionsTable 17: Testing and Validation PatternsTable 18: Cost Optimization StrategiesTable 19: Monitoring and ObservabilityTable 20: Streaming vs Batch Ingestion

Table 1: Core Architecture Layers

Layer	Example	Description
Bronze Layer (Raw Data)	`bronze.sales_raw` `ingestion_time: 2026-04-12`	• Landing zone for unprocessed data exactly as received from sources • append-only, immutable, schema-on-read • preserves complete audit trail and enables reprocessability.
Silver Layer (Cleaned Data)	`silver.sales_cleaned` `WHERE is_valid = true`	• Refined and validated data with deduplication, type casting, null handling, and standardization • enforces schema-on-write • serves as enterprise-wide cleaned source.
Gold Layer (Business-Ready)	`gold.sales_by_region_daily` `SUM(revenue) GROUP BY region, date`	• Aggregated, denormalized data optimized for analytics and BI • star schema models, KPIs, feature tables • read-optimized with fewer joins • consumption-ready for dashboards and ML.

Table 1: Core Architecture Layers

Layer	Example	Description
Bronze Layer (Raw Data)	`bronze.sales_raw` `ingestion_time: 2026-04-12`	• Landing zone for unprocessed data exactly as received from sources • append-only, immutable, schema-on-read • preserves complete audit trail and enables reprocessability.
Silver Layer (Cleaned Data)	`silver.sales_cleaned` `WHERE is_valid = true`	• Refined and validated data with deduplication, type casting, null handling, and standardization • enforces schema-on-write • serves as enterprise-wide cleaned source.
Gold Layer (Business-Ready)	`gold.sales_by_region_daily` `SUM(revenue) GROUP BY region, date`	• Aggregated, denormalized data optimized for analytics and BI • star schema models, KPIs, feature tables • read-optimized with fewer joins • consumption-ready for dashboards and ML.