
Monitoring Fabric with Capacity Metrics
Track capacity utilization and identify performance bottlenecks.
The Microsoft Fabric Capacity Metrics app provides visibility into how your Fabric capacity is being utilized, helping optimize performance and costs.
What is the Capacity Metrics App?
The Capacity Metrics app is a pre-built Power BI application that connects to your Fabric capacity and displays utilization metrics across all workloads.
Key Metrics
Capacity Units (CU) - Measure of compute consumption - Allocated based on SKU size - Shared across all Fabric workloads
Utilization - Percentage of capacity used - Breakdown by workload type - Peak vs average usage patterns
Throttling - When demand exceeds capacity - Impact on user experience - Frequency and duration
Setting Up Monitoring
Step 1: Install the App Search for "Microsoft Fabric Capacity Metrics" in AppSource and install it to your workspace.
Step 2: Connect to Capacity Provide your capacity ID and authenticate. The app will begin collecting metrics.
Step 3: Review Dashboards Explore the pre-built reports showing: - Overall utilization trends - Workload breakdown - Interactive vs background operations - Item-level consumption
Interpreting Results
High Utilization - Consider upgrading capacity - Optimize expensive operations - Schedule heavy jobs off-peak
Uneven Usage - Redistribute workloads across time - Consider autoscale options - Review query efficiency
Optimization Actions
- Pause development capacities when not in use
- Schedule large refreshes during off-hours
- Optimize inefficient DAX queries
- Use incremental refresh to reduce load
- Consider dedicated capacities for critical workloads
Frequently Asked Questions
How often does the Capacity Metrics app refresh?
The app refreshes every 30 minutes by default. You can see real-time utilization in the Azure portal, but the app provides more detailed historical analysis.
Can I set up alerts for capacity issues?
Yes, you can create Power BI data alerts on key metrics like utilization percentage or throttling events to receive notifications when thresholds are exceeded.