If you haven’t already, check out Arun Ulag’s hero blog “FabCon and SQLCon 2026: Unifying databases and Fabric on a single, complete platform” for a complete look at all of our FabCon and SQLCon announcements across both Fabric and our database offerings.
Mirroring in Microsoft Fabric was built to remove the friction of traditional data integration. With a simple, zero‑ETL experience, customers can continuously replicate data from operational systems directly into OneLake—making that data instantly available for analytics, AI, and reporting across Fabric.
As more customers adopt Mirroring at scale, a consistent theme has emerged: replication alone isn’t always enough.
Customers want:
- Faster freshness without full reloads
- Incremental processing for downstream pipelines
- Business‑ready data that already reflects source‑defined logic
To meet these needs, we’re introducing Extended Capabilities in Mirroring — optional, paid enhancements that build on top of core Mirroring to unlock more advanced, real‑world analytics scenarios. These extended capabilities help customers move beyond replication to incremental analytics and curated data experiences in OneLake.
Core Mirroring vs. Extended Capabilities
At its foundation, core Mirroring remains free and unchanged.
Core Mirroring includes:
- Continuous replication into OneLake
- Automatic conversion to open Delta format
- Built‑in SQL analytics endpoints
- Analytics‑ready data accessible across Spark, SQL, Power BI, and Fabric workloads
Extended Capabilities are opt‑in enhancements designed for customers who need more granular change tracking or want to preserve business logic from their source systems.
Why Extended Capabilities Are Charged
Extended Capabilities introduce additional continuous processing beyond basic replication.
Features like CDF and Views require ongoing change detection, transformation, and emission of data—work that goes beyond simply keeping tables in sync. To reflect this incremental value, Extended Capabilities are billed using Fabric’s capacity‑based consumption model, aligned with the rest of the platform.
This approach ensures:
- Charges apply only when extended features are enabled
- Billing is tied to actual processing activity, not idle state
- A single, unified model across all extended mirroring features
- Transparency and predictability for customers at scale
Core Mirroring remains free; Extended Capabilities are optional and value‑driven. To learn more about extended capabiliting in Mirroring are charged, please reference our Extended Capabilties in Mirroring documentation.
Delta Change Data Feed (CDF): Incremental Changes, Continuously Applied
Many analytics and data engineering scenarios don’t need full table refreshes—they need to know what changed, and when.
Delta change data feed (CDF) captures inserts, updates, and deletes at a granular level and applies them incrementally into OneLake. Instead of reprocessing entire datasets, downstream consumers can work with change‑only data, enabling lower latency and more efficient pipelines.
With CDF, customers can:
- Process incremental changes in near real‑time
- Uses change-only processing
- Enable downstream incremental analytics and AI workflows
- Eliminate custom change‑tracking pipelines
CDF works seamlessly across all mirroring sources—including Oracle, Snowflake, Azure SQL, and Open Mirroring partners—using native Delta semantics in OneLake. More details about how to set up and get started can be found in the Overview of Delta Change Delta Feed documentation.
In addition to batch consumption using Fabric Spark, the Delta change feeds can also be streamed in real-time. Using the new Mirrored Database Change Feed connector (Preview), users can publish CDF changes directly into Fabric Eventstreams, enabling low-latency, event-driven applications and real-time analytics using Real-Time Intelligence components like Eventhouse & Activator. To learn more about this capability, please reference the CDF with Eventstreams documentation.
Mirroring Views: Bringing Business Logic Directly into OneLake
Analytics teams don’t consume raw tables—they consume curated datasets shaped by business logic.
Mirroring views allow customers to replicate logical views from their source systems directly into OneLake, preserving filters, joins, and transformations defined upstream. Instead of rebuilding logic with ETL pipelines, Fabric mirrors the results as Delta tables, ready for analytics and AI workloads.
With mirroring views, customers can:
- Replicate source‑defined views instead of physical tables
- Reduce duplication and transformation complexity
- Ensure consistent business definitions across systems
- Make curated data immediately available in Fabric
Today, views are available in preview for Mirroring for Snowflake, with additional sources coming soon. More details about how to get starting with replicating views in Snowflake can be found in the Overview of Snowflake views documentation.
Designed for Real‑World Analytics and AI - Get Started
Together, delta change data feed and mirroring views represent a shift in how customers use Mirroring—not just to move data, but to power smarter analytics foundations.
Customers are using Extended Capabilities to:
- Keep dashboards continuously fresh with minimal compute
- Drive incremental data pipelines without custom ETL
- Preserve business logic across cloud and on‑prem systems
- Enable faster experimentation and AI‑ready data products
Extended Capabilities are available today in preview and can be enabled during mirror setup in the Fabric workspace. To learn more:
- Review Extended Capabilities in Mirroring – Overview
- Explore delta change data feed and views in the Fabric documentation
- Try Extended Capabilities in your existing mirrored databases
We look forward to seeing how customers use these capabilities to unlock even more value from their data in Microsoft Fabric.