Topic 03 · DP-700 · DP-600

Eventhouse, KQL, and Real-Time Intelligence

Learn how Fabric receives streaming events, stores them for fast operational analysis, and turns conditions in live data into actions.

The real-time mental model

Eventstreams move and shape events, Eventhouse stores them, KQL queries them, dashboards show the current state, and Activator responds when a rule is satisfied.

Eventstreams: route data as it arrives

Use Eventstreams to capture events from sources such as Event Hubs, IoT Hub, Kafka, or Fabric events. Filter, project, aggregate, join, and window events before sending them to an Eventhouse, Lakehouse, Activator, or another destination.

Choose Eventstreams when data needs transformation or routing while it is in motion.

Eventhouse and KQL databases

An Eventhouse contains KQL databases designed for append-heavy, time-series and operational workloads. Data is queried with Kusto Query Language (KQL), and the same data can feed dashboards, Power BI, and automated actions.

Choose an Eventhouse for high-volume event data that is queried by time, entity, or operational state.

Write efficient KQL

Apply restrictive where filters early, especially time filters. Use project to return only the columns needed, put the smaller input on the left side of a join, and use limit while exploring large results.

Reduce the amount of data scanned before expensive joins, projections, or aggregations.

Dashboards and Activator

Real-Time Dashboards display refreshed KQL query results. Activator follows a Connect, Monitor, Act pattern: it watches properties of real-world objects and triggers actions when a rule condition is met.

Use Activator when a condition should cause an action, not merely appear in a report.

Which component should you choose?

Comparison of Fabric real-time intelligence components
Requirement	Best fit
Transform or route events before storage	Eventstream
Store and query high-volume time-series data	Eventhouse with a KQL database
Create reusable query logic	Stored KQL function
Precompute recurring aggregations	Materialized view
Trigger a notification or workflow from a condition	Activator

Exam checklist

• Filter by time early and project only needed columns to reduce query cost.
• Use a materialized view for a recurring aggregation over a large, changing dataset.
• Use a stored function when the same KQL logic must be reused or parameterized.
• Use a window in Eventstream or KQL when the business rule depends on a time interval.

Previous topicNext shared topic coming soon