Microsoft Fabric Lakehouse — Medallion Architecture

The medallion architecture in a Fabric Lakehouse organizes data into Bronze, Silver, and Gold layers. This pattern progressively structures raw ingested data through cleansing and enrichment into refined Delta Lake tables optimized for analytics.

You implement Bronze, Silver, and Gold pipelines by ingesting raw data into Bronze, cleansing and enforcing schemas in Silver, and delivering optimized analytics tables in Gold using Spark or SQL endpoints with Delta Lake on OneLake.

Yes, the Fabric Lakehouse medallion architecture supports Delta Lake MERGE upserts for data updates and V-Order optimization for analytics performance, ensuring efficient querying and data management across your pipeline layers.

Yes, Fabric supports cross-workspace data sharing using OneLake Shortcuts. This allows you to reference and share Delta Lake tables across different workspaces without duplicating data across your medallion architecture layers.

You handle schema enforcement and data quality during the Silver layer transformation phase. This involves cleansing data, applying schema rules, deduplication, and PII masking before promoting validated records to the Gold analytics layer.

The best way to maintain Delta Lake tables is by running periodic OPTIMIZE and VACUUM maintenance tasks. These operations compact small files and remove obsolete data, keeping your Fabric Lakehouse medallion layers performant.