Database Mirroring in Microsoft Fabric

Fabric Mirroring enables near real-time replication of operational databases to OneLake, allowing analytics without impacting production systems.

What is Database Mirroring?

Mirroring continuously replicates data from transactional databases to Fabric using change data capture (CDC). The replicated data lands in OneLake in Delta format.

Supported Sources

Azure SQL Database
Azure Cosmos DB
Snowflake
(More sources being added)

How It Works

Initial Snapshot Full copy of selected tables to OneLake

Continuous Sync CDC captures changes and applies them - Inserts, updates, deletes tracked - Near real-time latency (seconds to minutes) - Minimal impact on source system

Delta Format Replicated data stored as Delta tables: - Queryable with SQL and Spark - Time travel capabilities - Automatic optimization

Setting Up Mirroring

Step 1: Create Mirrored Database In your Fabric workspace, select new Mirrored Database and choose source type.

Step 2: Configure Connection Provide source database credentials and connection details.

Step 3: Select Tables Choose which tables to replicate. Consider: - Tables needed for analytics - Change frequency - Data volume

Step 4: Start Mirroring Initiate the process. Initial sync may take time for large tables.

Use Cases

Real-time operational dashboards
Customer 360 views
Fraud detection
Inventory monitoring
Cross-database analytics

Frequently Asked Questions

Is database mirroring real-time?

Mirroring is near real-time, typically with latency measured in seconds to a few minutes depending on change volume and source database performance. It uses CDC for continuous synchronization.

Does mirroring affect my production database?

Impact is minimal. Mirroring uses CDC which has low overhead. For Azure SQL, it uses built-in change tracking. Always monitor source performance during initial setup.