Materialized Lake Views in Microsoft Fabric: When Your Medallion Fits in a SELECT Statement

2026-06-20 · Source: Towards Data Science · Field: Technology & Digital — Cloud Computing & IT Infrastructure, Data Science & Analytics · Depth: Intermediate, long

Summary

Materialized Lake Views (MLVs) in Microsoft Fabric, which became generally available at FabCon Atlanta in March 2026, simplify medallion architecture by enabling entire data pipelines to be defined using declarative Spark SQL or PySpark. MLVs are persisted, automatically refreshed views stored as Delta tables, handling execution, storage, refresh, dependency tracking, and data quality. Key enhancements since their Build 2025 preview include multi-schedule support, broader incremental refresh coverage for aggregations and joins, PySpark authoring (currently in preview with full refresh only), in-place updates via "CREATE OR REPLACE", and stronger expression-based data quality constraints. MLVs automatically infer dependencies, providing visual lineage and central error reporting, significantly reducing the complexity of traditional notebook-and-pipeline orchestrations. They require a schema-enabled lakehouse and Fabric Runtime 1.3 or higher.

Key takeaway

For Data Engineers or MLOps Engineers building medallion architectures in Microsoft Fabric, Materialized Lake Views offer a compelling alternative to complex notebook-and-pipeline orchestrations. If your transformations are primarily SQL-based, consider adopting MLVs to consolidate logic, automate refresh schedules, and embed data quality checks directly. This approach simplifies debugging and maintenance, allowing you to define "what" your data layers should look like, rather than "how" they are built. Ensure source tables have Delta change data feed enabled for optimal incremental refresh performance.

Key insights

Materialized Lake Views in Microsoft Fabric streamline medallion architectures by consolidating data transformation, refresh, and quality into declarative SQL or PySpark.

Principles

• Declarative definitions simplify complex data pipelines.
• Automatic dependency tracking enhances reliability.
• Incremental refresh optimizes compute costs.

Method

Define transformations using Spark SQL "CREATE MATERIALIZED LAKE VIEW" or PySpark "@fmlv.materialized_lake_view" decorator. Enable Delta change data feed on source tables for optimal incremental refresh.

In practice

• Use MLVs for frequently accessed aggregations or complex joins.
• Implement expression-based data quality constraints for uniform validation.
• Leverage "CREATE OR REPLACE" for in-place MLV definition updates.

Topics

• Microsoft Fabric
• Materialized Lake Views
• Medallion Architecture
• Spark SQL
• PySpark
• Data Quality
• Delta Lake

Best for: Data Engineer, MLOps Engineer, AI Architect

Related on AIssential

• Understanding the Medallion Architecture: A Practical Approach to Building Reliable Data Platforms — Medallion Architecture refines raw data through distinct layers to improve quality, governance, and usability.
• Databricks at SIGMOD 2026 — Enzyme simplifies complex ETL workloads by incrementally maintaining Materialized Views across diverse data patterns and languages.
• Lakehouse vs Warehouse in Microsoft Fabric Practical Guide — The choice between Lakehouse and Warehouse in Microsoft Fabric hinges on real-world architectural needs, not just definitions.
• Backstage with Lakebase, part 2 — Unified governance for operational databases simplifies security, compliance, and cost attribution while enabling rapid development workflows.
• Building ETL Pipelines in Databricks | Data Engineering in Databricks — Databricks ETL pipelines transform raw data into production-ready insights using Medallion Architecture and AI-assisted coding.
• #355 AI's Impact on Databases with Shireesh Thota, CVP of Databases at Microsoft — Unified data platforms and AI agents are transforming data interaction, enabling deeper reasoning and simplified management.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Towards Data Science.