Advanced2 daysNew
Real-Time Analytics Architecture on Microsoft Fabric
Event-driven analytics at scale: Service Bus, Eventstream, Eventhouse, and Activator
Overview
Most organisations can handle batch analytics. The genuinely hard problem is acting on data within seconds of it being created — detecting fraud before a transaction completes, triggering a personalised offer at the moment a customer shows buying intent, or alerting a plant operator before a machine fails. This two-day architecture training teaches you to build production-grade, sub-second analytics systems on Microsoft Fabric using the event-driven architecture pattern published in the Azure Architecture Center. You will wire together Azure Service Bus for discrete transactional events, Fabric Eventstream for high-velocity continuous data, Fabric Eventhouse (KQL) as the analytical store, and Fabric Activator to trigger automated actions — then layer on real-time Power BI dashboards, Fabric Copilot, and data agents for AI-driven operational intelligence.
What you'll learn
- Design a dual-track event ingestion architecture: Service Bus for transactional events (orders, loyalty updates) and Eventstream for high-velocity streams (clickstream, telemetry)
- Build and operate a Fabric Eventhouse with KQL tables optimised for time-series queries, aggregations, and anomaly detection
- Configure Fabric Eventstream pipelines to fan out, filter, and transform events in-flight before landing in Eventhouse or Lakehouse
- Set up Fabric Activator rules to trigger automated responses — Teams notifications, Power Automate flows, Azure Function calls — based on real-time data conditions
- Build sub-second Power BI real-time dashboards connected to Eventhouse via DirectQuery for operational monitoring
- Integrate Microsoft Purview governance and Microsoft Entra-based security into a live event-driven analytics platform
Programme
Day 1 — Event ingestion, Eventstream & Eventhouse
- Real-time analytics architecture overview: where batch ends and event-driven begins, and why the choice matters for business outcomes
- Azure Service Bus for transactional events: topics, subscriptions, dead-letter queues, and the difference between Service Bus and Event Hubs for analytics use cases
- Fabric Eventstream: creating streams, connecting sources (Event Hubs, IoT Hub, Kafka, custom endpoints), and routing events to multiple destinations in parallel
- In-flight transformations in Eventstream: filtering, field projection, aggregation windows, and joining reference data to enrich events before storage
- Fabric Eventhouse and KQL tables: partitioning strategy, hot vs cold cache, retention policies, and continuous ingestion from Eventstream
- KQL for real-time analytics: time-series operators, summarize, make-series, anomaly detection with series_decompose_anomalies, and joining live streams against historical Lakehouse data
- Hands-on: build an end-to-end event pipeline from an Azure Service Bus topic through Eventstream into an Eventhouse, and write KQL queries to detect threshold breaches in real time
Day 2 — Activator, dashboards, AI & governance
- Fabric Activator: creating rules and reflexes that monitor Eventstream or Eventhouse data and fire automated actions without custom code
- Activator action targets: sending Teams cards, triggering Power Automate flows, calling Azure Functions, and writing back to operational systems
- Real-time Power BI dashboards on Eventhouse: DirectQuery vs streaming datasets, auto-refresh, and building sub-second operational monitoring views
- Fabric data agents and Copilot Studio integration: enabling natural-language queries against live Eventhouse data for non-technical stakeholders
- Layering historical context: mirroring or copying Eventhouse aggregates to Fabric Lakehouse for unified reporting across historical and real-time data
- Governance and security: Microsoft Purview integration with Fabric Real-Time Intelligence, sensitivity labels on event data, and workspace-level RBAC
- Production patterns: disaster recovery, geo-redundancy, capacity planning, and cost optimisation for high-throughput event workloads on Fabric
- Hands-on: configure Activator to trigger a Teams alert and a Power Automate flow from a live KQL rule, then publish a real-time Power BI dashboard for the same data stream
Who is this for?
- Data architects and platform engineers designing real-time or near-real-time analytics systems on Azure
- Data engineers migrating from Spark Streaming or Azure Stream Analytics to the Microsoft Fabric Real-Time Intelligence suite
- Analytics engineers who need to close the gap between data arrival and business action to under one second
- Solution architects in retail, manufacturing, financial services, or IoT verticals evaluating Fabric for operational analytics
Prerequisites
- Solid understanding of Azure data services (Event Hubs, Storage, Synapse or Fabric fundamentals)
- KQL basics or willingness to learn on day one — a primer is included
- Familiarity with streaming concepts: events, partitions, consumer groups, and windowing functions
Tools & technologies covered
Microsoft FabricFabric EventstreamFabric EventhouseFabric ActivatorAzure Service BusKQLPower BIFabric CopilotMicrosoft PurviewAzure Monitor
Not sure which course fits your team?
Talk to us — we'll match you to the right training path.