Skip to content

Mirror Databases in Microsoft Fabric

Mirror Databases in Microsoft Fabric

Source

Summary

This module explores Database Mirroring in Microsoft Fabric, which provides continuous, near-real-time replication of transactional operational databases into Delta Lake format inside OneLake without writing custom ETL pipelines.

Supported Sources & Mechanism

1. Supported Mirroring Sources

  • Azure SQL Database
  • Azure Cosmos DB (NoSQL)
  • Snowflake
  • Azure Database for PostgreSQL / MySQL

2. Inner Mechanics

  • Utilizes native source database change tracking / CDC logs to replicate commits asynchronously into OneLake.
  • Automatically writes changes in Delta Lake (Parquet) with V-Order applied.
  • Automatically provisions a read-only SQL Analytics Endpoint for direct cross-database T-SQL querying and Power BI Direct Lake reports.

3. Benefits over Traditional ETL

  • Zero complex ETL pipeline development.
  • Continuous replication with sub-minute latency.
  • No compute overhead on operational transactional source queries.

Exam Traps & Gotchas

[!IMPORTANT]

  • Prerequisites for Azure SQL Database Mirroring: System-assigned managed identity (SAMI) must be configured, and CDC / Change Tracking must be enabled on target tables.
  • Read-Only Target: Mirrored tables in OneLake are read-only; write mutations must occur at the upstream operational source database.