Change Event Streaming for SQL database in Microsoft Fabric (Preview)

Microsoft Employee

1 month ago

SQL database in Microsoft Fabric is a developer-friendly transactional database designed for operational workloads inside Fabric. It already replicates data to OneLake automatically, enabling analytics across your Fabric estate. But what if your downstream systems need to react to changes closer to when they happen, not several stacked transformation batches later?

Change Event Streaming (CES) streams row-level changes—inserts, updates, and deletes—from your Fabric SQL database directly to Fabric Eventstream and Azure Event Hubs in outbound, continuous sending fashion. This unlocks the possibility of building near real-time data pipelines on top of operational data without leaving Fabric.

A header image depicting events flowing from a Fabric SQL database to Fabric Eventstream and further branching into multiple consumers from there.

Figure: Change Event Streaming concept diagram.

From operational data to Real-Time Dashboards within Fabric

SQL database in Fabric already bridges operational and analytical workloads. Automatic mirroring into OneLake gives you a SQL analytics endpoint for reporting and cross-database queries. CES adds a complementary path: instead of waiting for data to land in OneLake and get transformed through polling and a number of batches, you can stream changes out upon commit, and route them through Fabric Eventstream for Real-time processing.

This completes a streaming SQL data continuum inside Fabric:

Fabric SQL database → CES → Fabric Eventstream → Real-Time Intelligence

Your operational data stays in Fabric. Your pipelines stay in Fabric. Your analytics stay in Fabric. From Fabric SQL to Fabric Eventstream scenario, you just connect existing Fabric components together.

What is Change Event Streaming (CES)?

Change Event Streaming captures committed row-level changes from your database using transaction log-based capture and publishes them as structured JSON events following the CloudEvents standard. Each event includes the operation type, primary key, and before/after values of changed columns.

CES is configured using T-SQL stored procedures directly in your database. You define a stream group that represents the destination and add tables to it. SQL handles the rest—tracking which changes have been delivered, retrying on transient failures, splitting large events to conform with destination limits.

Why CES for SQL database in Fabric?

Near-real-time analytics on operational data

Consider a scenario where your application writes an order, inventory, or customer data to a SQL database in Fabric. With CES, you can configure SQL to stream those changes directly into Fabric Eventstream.

From Eventstream, you can route them to an Eventhouse for Real-Time Dashboards, apply stream processing with SQL operators to avoid transforming data in batches, or trigger actions with Activator.

Example scenarios that benefit from more current data include inventory alerts, customer experience signals, operational monitoring, just to name a few.

A simpler architecture within Fabric

Without CES, getting data to stream out of the database typically means assembling a pipeline from CDC, Change tracking, polling jobs and connectors. Each component adds credentials to manage, failure points to monitor, and latency to tune.

With CES, Fabric SQL database pushes changes directly to Fabric Eventstreams. You don’t need to install components for the data to flow between these two systems.

Decoupled consumers

Once events arrive in Fabric Eventstream, downstream processes can consume them independently. A Fabric Real-Time Dashboard, a data pipeline into a lakehouse, and an Activator rule can all subscribe to the same stream. The consumers no longer interact directly with SQL - Eventstream handles the distribution.

Getting started

To configure CES on your SQL database in Fabric:

Locate or create a Fabric Eventstream in your workspace and add a custom endpoint source.
Retrieve the connection details from the custom endpoint.
In your SQL database, enable event streaming and create a stream group pointing to the Eventstream endpoint.
Add the tables you want to stream.

For step-by-step guidance, refer to the Stream SQL change events to Eventstream for real-time processing tutorial.

Fabric Eventstream editor showing incoming CES events from a SQL database