metricsign
Start free
Medium severitycapacity

Power BI Refresh Error:
Refresh operation throttled by Power BI Premium

What does this error mean?

Power BI Premium throttles data refresh operations when too many semantic models are being processed concurrently on a capacity, causing refreshes to be delayed or queued. This is a capacity-level constraint, not a model-level error.

Common causes

  • 1Too many semantic models scheduled to refresh simultaneously on the same Premium capacity
  • 2Large or complex semantic models consuming disproportionate capacity resources and starving concurrent refreshes
  • 3Refresh schedules bunched at peak hours (e.g., top of the hour) creating concurrency spikes
  • 4Undersized Premium capacity SKU relative to the number of models and refresh frequency required

How to fix it

  1. 1Step 1: Review all scheduled refresh times across semantic models on the affected capacity and stagger them to distribute load — avoid scheduling multiple large models at the same time.
  2. 2Step 2: Identify the largest or most resource-intensive semantic models using the Premium Capacity Metrics app and consider splitting them into smaller semantic models.
  3. 3Step 3: Move non-critical or lower-priority semantic models to a different capacity or workspace to reduce concurrency on the affected node.
  4. 4Step 4: If throttling is persistent, evaluate upgrading to a larger Premium SKU or enabling autoscale if using Premium Per User or Fabric capacity.
  5. 5Step 5: Retry the throttled refresh operation during off-peak hours or reschedule it to a time with lower concurrent activity.

Frequently asked questions

How many concurrent refreshes does Power BI Premium support?

The maximum number of concurrent refreshes depends on the Premium SKU. For example, P1 supports up to 6 concurrent refreshes. Exceeding this limit causes additional refresh requests to be queued or throttled.

Does retrying the refresh immediately help when throttled?

A retry may succeed if other models have finished refreshing and capacity is freed up, but repeated immediate retries during a peak load period are likely to be throttled again. Scheduling the retry for an off-peak window is more effective.

Other capacity errors