End-to-End Fabric Monitoring: Real-Time Signals, Logs, and Unified Observability Patterns

Monitoring Microsoft Fabric effectively requires combining real-time signals, operational events, and logs into a unified observability strategy. This session walks through how administrators, engineers, and platform owners can monitor Fabric in near real-time using Capacity Events, Spark Monitoring, Activity Events, and workspace monitoring.
We begin by exploring Capacity Overview Events and Capacity Operation Events, which surface CU usage, interactive delay thresholds, and capacity health through the Real-Time Hub. You’ll learn how these events can be consumed through Eventstream, processed with Activator, and stored for deeper analytics in Eventhouse or Lakehouse.
Next, we dive into Spark Real-Time Monitoring, based on the integration patterns with the Spark Diagnostic Emitters, explaining how Spark job definitions and notebooks can send operations execution signals to the Eventstream.
The session then covers Activity Events in smaller batches, enabling lower-latency extraction and faster troubleshooting cycles, and ties these into the broader platform monitoring surface illustrated in Fabric Platform Monitoring.
Finally, we show how to unify everything with Workspace Monitoring through Eventhouse, and how to design a hybrid monitoring pattern that correlates events across workspace, capacity and tenant.
By the end of the session, attendees will understand how to combine all monitoring entry points—Real Time Hub, Spark monitoring signals, capacity events, and workspace analytics—into a single, coherent observability layer for Fabric.

- How to combine real-time and log-based signals into a unified Fabric monitoring architecture - Practical patterns for using Capacity Events, Spark signals, and Activity Events together - Techniques to correlate events across workloads for faster troubleshooting and insights