Monitor and optimize an analytics solution Study Guide — DP-700

Monitor and Optimize an Analytics Solution

Microsoft publishes all three DP-700 domains at 30–35% each (not exact point values) — this domain is roughly tied with the other two, not the single largest. It tests your ability to monitor Microsoft Fabric workloads end-to-end, configure proactive alerting, diagnose and fix errors across pipelines, dataflows, notebooks, eventhouses, and eventstreams, and apply performance-tuning techniques at both the storage and compute layers. Expect scenario-based questions that require you to choose the correct monitoring tool, optimization command, or alert configuration for a specific Fabric item type.

Monitoring Data Ingestion, Transformation, and Semantic Model Refresh

Core Monitoring Tools

Monitoring Hub Essentials

The Monitoring Hub is the unified entry point for all activity monitoring in Fabric. Key behaviors:

• Shows runs for pipelines, dataflows, notebooks, Spark job definitions, semantic model refreshes, and lakehouses.
• Retains history for 30 days by default.
• Supports filtering by item type, status (Succeeded/Failed/In Progress/Cancelled), and time range.
• You can re-run a failed pipeline or dataflow directly from the Monitoring Hub without navigating to the item.

Semantic Model Refresh

• Scheduled refresh is configured in the semantic model settings (up to 48 refreshes per day on Premium/Fabric capacity).
• Import mode models require a full or incremental refresh; DirectLake models do not require a traditional refresh — they read from the Delta table directly.
• Refresh failures surface in both the Monitoring Hub and the semantic model's own Refresh history pane with error codes.
• Gateway-dependent refreshes (on-premises data sources) add gateway availability as a failure vector.

Common Traps

• Confusing Dataflow Gen1 (Power Query Online, different refresh path) with Dataflow Gen2 (Fabric, monitored in Monitoring Hub).
• Assuming DirectLake models need scheduled refresh — they do not unless you want to update the model schema itself.
• Monitoring Hub history is 30 days, not unlimited; for longer retention you must export to a Log Analytics workspace or use Fabric Capacity Metrics app.

Configuring Alerts

Alert Types and Where to Configure Them

Data Activator (Reflex)

Data Activator is Fabric's native no-code alerting engine. Key exam facts:

• Receives data from Eventstream, Power BI reports (visual-level), or Real-Time hub.
• Triggers actions: Teams notification, email, Power Automate flow, or Fabric item run.
• Conditions: value crosses threshold, enters/exits a range, changes by a percentage, or no data received within a time window (data absence alert).
• Each alert definition is a Reflex item stored in a Fabric workspace.

Setting Up Pipeline / Dataflow Alerts

• In Monitoring Hub, selecting a completed/failed run exposes a "Set alert" option.
• Alerts can notify via Teams or email.
• For programmatic alerting, export logs to Azure Monitor / Log Analytics and create Azure Monitor alert rules.

Common Traps

• Data Activator alerts on streaming data conditions — it is NOT the right tool to alert on a pipeline run failure (use Monitoring Hub alert or Azure Monitor for that).
• Confusing Reflex (the Data Activator item type) with a generic webhook — Reflex is a first-class Fabric item.
• Visual-level alerts from Power BI reports require the report to be open in a browser session unless connected through Data Activator via Real-Time hub.

Worked scenario: A requirement states "alert the operations team in Microsoft Teams when the value of a streaming sensor reading exceeds 95°C." The correct answer is Data Activator connected to an Eventstream, not a pipeline alert or Azure Monitor metric alert, because the condition is on live streaming data values.