DataTerminology.comNEW
A layered lakehouse pattern organizing data into Bronze (raw), Silver (cleaned), and Gold (business-ready) tiers.
DESCRIPTION
Medallion architecture structures data pipelines into progressive quality tiers. Bronze captures raw ingested data; Silver standardizes and de-duplicates; Gold delivers business-ready aggregates and marts. It improves governance, lineage, and reliability while enabling incremental development. Often implemented with Delta Lake, orchestration tools, and quality checks per layer. It suits both batch and streaming ingestion and clarifies ownership across the refinement path.
ALTERNATIVE DEFINITIONS
Databricks
The medallion architecture is a data design pattern to organize data in a lakehouse, with Bronze, Silver, and Gold layers.
Defines the three-tier pattern within a lakehouse context.
Microsoft Learn
The medallion architecture organizes data into Bronze (raw), Silver (cleaned), and Gold (business) layers.
Uses explicit semantics for each tier’s purpose.
RELATED TERMS
Data Lakehouse Delta Lake Bronze Silver Gold Extract-Transform-Load ELT Data Quality Lineage Curated Layer Staging Orchestration
COMPETING TERMS
Lambda Architecture Kappa Architecture
KNOWLEDGE AREAS
Data Architecture Data Engineering Data Warehousing & BI Real-time & Streaming Data
TARGET AUDIENCE
Data Engineer Data Architect Analytics Engineer BI Specialist MLOps Engineer Cloud Architect
EXTERNAL SOURCES
- Databricks - Implementation guidance for medallion tiers
- Microsoft - Medallion design in Microsoft Fabric
TOP QUESTIONS
Why use Medallion Architecture?
It standardizes data refinement, improves quality and lineage, and enables incremental delivery from raw to business-ready datasets.
Category: Value
What belongs in the Bronze layer of Medallion Architecture?
Raw, ingested data in its original structure with minimal transformation, primarily for traceability and reprocessing.
Category: Concepts
How is the Silver layer defined in Medallion Architecture?
Silver contains cleaned, standardized, de-duplicated, and conformed data suitable for joining across domains.
Category: Concepts
What is the Gold layer used for in Medallion Architecture?
Gold delivers business-ready models, aggregates, and marts for BI, analytics, and ML features.
Category: Concepts
How does Medallion Architecture support streaming?
Streaming jobs can populate Bronze continuously, with incremental upserts to Silver and Gold using merge and watermarking.
Category: Implementation
What are common errors with Medallion Architecture?
Skipping data quality in Bronze-to-Silver, unclear ownership at each tier, and tight coupling across layers.
Category: Mistakes
How to govern Medallion layers?
Define data contracts, quality SLAs, lineage tracking, and access controls per layer with clear custodians.
Category: Governance
Does Medallion Architecture replace a Data Warehouse?
It can complement or underpin warehouses; Gold often materializes star schemas or semantic models.
Category: Opinion
How to measure success of Medallion Architecture?
Track freshness, quality metrics, downtime, consumer adoption, and cycle time from raw to production datasets.
Category: Metrics