Introduction
Managing capacity in Microsoft Fabric goes beyond provisioning the right SKU — it’s about understanding how that capacity behaves under real workloads.
Fabric’s Capacity Metrics App is designed for exactly that: to give administrators and platform owners visibility into compute consumption, workload behavior, and system health across every experience running under a shared capacity.
Think of it as your “flight dashboard” for Fabric — surfacing not just how much capacity you’re using, but where, when, and why it’s being consumed.
What the Metrics App Is
The Microsoft Fabric Capacity Metrics App is a prebuilt Power BI report developed by Microsoft, specifically for monitoring Fabric capacities. It connects directly to your tenant’s telemetry and surfaces insights such as:
- How your capacity’s compute resources are being used.
- Which workloads and experiences (e.g., Data Engineering, Data Warehouse, Power BI) consume the most CUs.
- When your capacity experiences throttling or queueing events.
- Trends in user activity and job execution over time.
The app transforms raw telemetry into a unified operational view — enabling data leaders to optimize performance, plan scaling, and detect cost inefficiencies early.
Installing the Metrics App
You can install the Metrics App directly from the Microsoft AppSource or the Fabric admin portal.
To install the Microsoft Fabric Capacity Metrics app for the first time, follow these steps:
- Select one of these options to get the app from AppSource:
- Go to AppSource > Microsoft Fabric Capacity Metrics and select Get it now.
- In Power BI service:
- Select Apps.
- Select Get apps.
- Search for Microsoft Fabric.
- Select the Microsoft Fabric Capacity Metrics app.
- Select Get it now.
- When prompted, sign in to AppSource using your Microsoft account and complete the registration screen. The app takes you to Microsoft Fabric to complete the process. Select Install to continue.
- In the Install this Power BI app window, select Install.
- Wait a few seconds for the app to install.
Once installed, the app automatically connects to your Fabric telemetry and begins pulling in capacity metrics. It refreshes periodically, ensuring admins always have access to near-real-time usage data.
(Reference: Official Installation Guide)
How to Read the Metrics App
The Metrics App contains several pages, each focused on a specific dimension of capacity behavior.
Let’s break down the key ones:
1. Overview Page — The Health Snapshot
The Overview tab gives a high-level summary of capacity performance, covering:
- CPU and Memory Utilization: Shows average and peak compute usage across time.
- Throttling Incidents: Counts and duration of capacity throttling events.
- Queue Lengths: Indicates whether workloads are being delayed due to high demand.
- Autoscale Activity: Displays when autoscaling occurred, highlighting demand spikes.
This page answers the fundamental question: “Is my capacity healthy right now?”
2. Compute Page — Deep Dive into Usage
The Compute page is the heart of the Metrics App. It provides granular visibility into how CUs are being consumed.
Key visuals include:
- Total Compute Usage Over Time: Identifies peak hours and workload surges.
- Experience-Level Breakdown: Splits usage across Data Engineering, Power BI, Data Warehouse, and Real-Time Analytics.
- Workload Type Distribution: Shows the proportion of interactive vs. background jobs.
- Throttling Timeline: Highlights when throttling occurred and which workloads contributed most.
This view helps you diagnose imbalance — for example, spotting that nightly ETL jobs are monopolizing capacity, or that Power BI refreshes spike during business hours.
(Reference: Compute Page Details)
3. Operations Page — Understanding Queueing and Job Behavior
The Operations page focuses on the throughput and concurrency of jobs. It helps answer:
- How many jobs are running concurrently?
- Are there delays or backlogs forming?
- Are specific workloads consistently queuing?
If you’re observing user complaints of “Fabric being slow,” this page reveals whether it’s genuine throttling or workload congestion from poor scheduling.
4. Users and Workspaces — Attribution and Accountability
Capacity transparency is meaningless without ownership.
These pages let you see who and which workspace is consuming the most capacity, supporting governance and cost attribution.
Common use cases:
- Identifying “noisy neighbors” in shared capacities.
- Justifying the need for dedicated capacity for certain teams.
- Tracking consumption patterns by department or project.
Using the Metrics App for Proactive Management
Having visibility is one thing; acting on it is another.
Here’s how platform owners can use the Metrics App to keep Fabric predictable and efficient:
- Detect Throttling Early
Regularly check the throttling timeline — consistent spikes at similar times often mean workload overlap or under-provisioned capacity. (Read More: Introduction to Fabric Capacity Management) - Optimize Scheduling
Use concurrency and job timing data to stagger heavy workloads (e.g., move ETL to off-hours). - Right-Size Capacity
Monitor sustained high utilization over weeks — a sign you’ve outgrown your SKU. - Enforce Governance
Use user and workspace insights to introduce quotas or realignment. - Plan Scaling Intelligently
Instead of reactive scaling after incidents, use trends from the Metrics App to forecast growth 1–3 months out.
Integrating with Broader Monitoring
While the Metrics App is the primary tool for capacity visibility, it works best as part of a broader observability strategy:
- Fabric Activity Logs for detailed audit and operational context.
- Power BI Admin Portal for model refresh telemetry.
- Azure Monitor / Log Analytics integration for central alerting and cross-platform insights.
Together, these layers help build a full operational view — not just of Fabric, but of how it interacts with the rest of your data estate.
Conclusion
The Fabric Capacity Metrics App transforms capacity management from reactive firefighting into proactive governance. It gives data leaders the visibility needed to detect throttling before it hurts performance, balance workloads intelligently, and ensure every compute unit is spent where it drives business value.
But visibility is just the beginning. The real advantage comes when capacity management and data architecture work together — when every job, dataset, and model runs efficiently because the foundation beneath it is structured right.
If you’d like to see how these principles connect in practice, join our upcoming live session:
👉 Capacity Management and Medallion Architecture on Fabric
We’ll explore how to bring the Medallion model and capacity optimization together to make Fabric environments faster, more cost-efficient, and ready for AI workloads.
Leave a Reply