Schema-aware eventstreams in Microsoft Fabric (preview)

Modern streaming solutions often process events with different, evolving, or drifting schemas and multiple event types in one pipeline. Schema-aware Eventstreams provide a unified event processing experience for both schematized and unschematized events. You can ingest, understand, and process events whether they arrive as semi-structured payloads or are tagged with registered schemas. This experience simplifies event discovery, exploration, and processing for heterogeneous event streams.

Important

Schema-aware Eventstreams are in Preview. When you create a new Eventstream, you choose whether to use the existing Eventstream experience or the schema-aware experience. You can't enable the schema-aware experience on an existing Eventstream. Existing Eventstreams continue to work as before.

Two ways to build an eventstream

When you create a new Eventstream, you can choose between two experiences:

• Existing Eventstream experience. Use the Eventstream capabilities available today. Existing Eventstreams and their features are unchanged.
• Schema-aware Eventstream (Preview). Opt in to process schematized, classified, and untyped events together in one Eventstream.

For the existing create flow, see Create an Eventstream in Microsoft Fabric. To opt in, see Create a schema-aware Eventstream (Preview).

Key capabilities

Explore streaming data with enhanced data preview

The redesigned data preview experience helps you inspect and understand streaming data before you build processing logic. You can:

• Explore deeply nested event payloads.
• View event headers and metadata.
• Export event details for further analysis.

For more information, see Preview data in a schema-aware Eventstream (Preview).

Classify unschematized events with different shapes

A single stream can contain unschematized events with different shapes. The Classifier workflow lets you identify a classifier or discriminator field, such as deviceType. Eventstream uses that field to distinguish and organize the event types while keeping them in one pipeline.

For more information, see Classify unschematized events (Preview).

Recognize events registered with Event SchemaSet

Schema-aware Eventstreams integrate with Event SchemaSet (GA) and the Fabric tenant-level schema registry. Events tagged with registered schemas can be automatically recognized and processed by using their corresponding schema definitions. This integration helps you govern, discover, and reuse data contracts across your organization.

Schematized events use the CloudEvents format. The ce-type header identifies the registered event type so Eventstream can apply the corresponding schema. Schema association and header requirements are documented with the supported source and Event SchemaSet workflows.

For more information, see Use event schemas in Eventstreams.

Process mixed event types in one Eventstream

Schematized events, classified events, and untyped events can coexist in the same Eventstream. Updated operators and destinations process these event types without requiring separate pipelines.

When you configure a no-code operator or a supported destination, use the Input schema list to choose a schematized event, a classified event, or the untyped/default input. The selection determines the event shape available to that operator or destination.

Use this capability when sources have different levels of schema maturity or when you want to adopt registered schemas incrementally.

When to choose schema-aware Eventstreams

Choose a schema-aware Eventstream when you need to:

• Process schematized and unschematized events in one pipeline.
• Inspect nested payloads, event headers, and metadata.
• Distinguish multiple unschematized event shapes by using a classifier or discriminator field.
• Recognize events that reference schemas registered with Event SchemaSet.
• Process schematized, classified, and untyped events with the same operators and destinations.

Use the existing Eventstream experience when you don't need these preview capabilities. Existing Eventstream features continue to work as before.

Preview considerations

• Schema-aware Eventstreams are opt-in when you create a new Eventstream.
• You can't convert an existing Eventstream to a schema-aware Eventstream.
• Existing Eventstreams aren't changed by this preview.
• Preview capabilities are subject to change before general availability.

Get started

1. Create a schema-aware Eventstream (Preview).
2. Preview data in a schema-aware Eventstream (Preview).
3. If your unschematized events have different shapes, classify the events.
4. To use registered schemas, follow the schema-association procedure for your supported source.