Skip to content

Ingest Real-Time Data with Eventstreams in Microsoft Fabric

Ingest Real-Time Data with Eventstreams in Microsoft Fabric

Source

Summary

This module covers configuring and routing streaming data with Microsoft Fabric Eventstreams, applying no-code event processing transformations (filtering, aggregations, windowing), and landing streams into multiple destinations.

Streaming Sources & Transformations

1. Supported Eventstream Sources

  • Custom App: Push events via REST API or Kafka protocol endpoints.
  • Azure Event Hubs & Azure IoT Hub: Direct connection to enterprise telemetry event brokers.
  • Amazon Kinesis Data Streams: Multi-cloud streaming ingestion.
  • Azure SQL Database CDC: Streaming change-data-capture logs.

2. In-Stream Processing Operators

  • Filter: Drop irrelevant events before landing.
  • Manage Fields: Rename, cast, or drop JSON attributes.
  • Aggregate: Tumbling, hopping, and sliding window aggregations over streaming time windows.
  • Union & Expand: Flatten nested JSON arrays and combine event streams.

3. Eventstream Destinations

  • KQL Database / Eventhouse: Sub-second indexing for real-time querying.
  • Lakehouse: Land stream directly into Delta tables in OneLake (Bronze layer).
  • Custom Endpoint / Reflex: Trigger automated alerts or actions based on streaming conditions via Data Activator.

Exam Traps & Gotchas

[!IMPORTANT]

  • Enhanced Capabilities Mode: When creating Eventstreams, ensure Enhanced Capabilities is enabled to support multiple destinations and rich stream transformations.
  • Direct Ingestion to Lakehouse: Eventstreams can write directly to Lakehouse Delta tables without requiring an intermediary Spark notebook.