Surge protection

2025-06-05

Surge protection helps limit overuse of your capacity by limiting the amount of compute consumed by background jobs. You configure surge protection for each capacity. Surge protection helps prevent throttling and rejections but isn't a substitute for capacity optimization, scaling up, and scaling out. When the capacity reaches its compute limit, it experiences interactive delays, interactive rejections, or all rejections even when surge protection is enabled.

Prerequisites

You need to be an admin on the capacity.

Surge protection thresholds

Capacity admins set a background operations rejection threshold and a background operations recovery threshold when they enable surge protection.

The Background operations rejection threshold determines when surge protection becomes active. The threshold applies to the 24-hour background percentage for the capacity. When the threshold is reached or exceeded, surge protection becomes active. When surge protection is active, the capacity rejects new background operations. When surge protection isn't enabled, the 24-hour background percentage is allowed to reach 100% before the capacity rejects new background operations.
The Background operations recovery threshold determines when surge protection stops being active. Surge protection stops being active when the 24-hour background percentage drops below the background recovery threshold. The capacity starts to accept new background operations.

Note

Capacity admins can see the 24-hour background percent on the Microsoft Fabric Capacity Metrics app Compute page under Throttling on the Background rejection chart.

Enable surge protection

To enable surge protection, follow these steps:

Open the Fabric Admin Portal.
Navigate to Capacity settings.
Select a capacity.
Expand Surge protection.
Set Background Operations to On.
Set a Rejection threshold.
Set a Recovery threshold.
Select Apply.

Monitor surge protection

To monitor surge protection, follow these steps:

Open the Microsoft Fabric Capacity Metrics app.
On the Compute page, select System events.

The System events table shows when surge protection became active and when the capacity returned to a not overloaded state.

Set the background operation rejection and recovery thresholds

Capacity admins need to evaluate the capacity's usage patterns when setting the rejection and recovery thresholds. Use the Microsoft Fabric Capacity Metrics app to evaluate the usage. On the Compute page, review the data in the Background rejection, Interactive rejection, and Utilization charts.

Example scenarios:

The Background rejection chart shows an average background percentage of 50%. The chart has several peaks at 60% and 75%. The lowest point on the chart is 40%. The Interactive rejection chart shows it exceeded 100% at the same time as the Background rejection chart peaked at 75%. To protect interactive users, setting a rejection threshold above 60% and below 75% would be an initial starting point. A recovery threshold above 40% and below 60% would be an initial starting point.
The Background throttling chart shows an average background percentage of 35%, and it usually varies by no more than 5%. The Interactive rejection chart shows a peak value of 80%, which means that interactive rejections aren't occurring. Setting a rejection threshold slightly above 40% and below 60% would be an initial starting point. Using a lower value would reduce the risk of impact to interactive users due to a surge in background operations. Setting the recovery threshold to 35% or even 40% is acceptable, because this value reflects the typical background utilization, and the capacity operates well with this level of usage.
The Utilization chart shows 80% or 90% of usage is from background operations; enabling surge protection background operation limits may not be helpful.

System events for surge protection

When surge protection is active, capacity state events are generated. The System events table in the Microsoft Fabric Capacity Metrics app shows the events. Below are the state events relevant to surge protection. A complete list of capacity state events is available in Understanding the Microsoft Fabric Capacity Metrics app compute page.

Capacity state	Capacity state change reason	When shown
Active	NotOverloaded	Indicates the capacity is below all throttling and surge protection thresholds.
Overloaded	SurgeProtectionActive	Indicates the capacity exceeded the configured surge protection threshold. The capacity is above the configured recovery threshold. Background operations are being rejected.
Overloaded	InteractiveDelayAndSurgeProtectionActive	Indicates the capacity exceeded the interactive delay throttling limit and the configured surge protection threshold. The capacity is above the configured recovery threshold. Background operations are being rejected. Interactive operations are experiencing delays.
Overloaded	InteractiveRejectedAndSurgeProtectionActive	Indicates the capacity exceeded the interactive rejection throttling limit and the configured surge protection threshold. The capacity is above the configured recovery threshold. Background and interactive operations are being rejected.
Overloaded	AllRejected	Indicates the capacity exceeded the background rejection limit. Background and interactive operations are being rejected.

Note

When the capacity reaches its compute limit, it experiences interactive delays, interactive rejections, or all rejections even when surge protection is enabled.

Per-operation status messages for surge protection

When surge protection is active, background requests are rejected. In the Fabric capacity metrics app, these requests appear with status Rejected or RejectedSurgeProtection. These status messages appear in the Microsoft Fabric Capacity Metrics Timepoint page. For more information, see Understand the metrics app timepoint page.

Considerations and limitations

When surge protection is active, background jobs are rejected. This means there's still broad impact across your capacity even when surge protection is enabled. By using surge protection, you're tuning your capacity to stay within a specific range of usage. However, while surge protection is enabled, background operations might be rejected, and this can impact performance. To fully protect critical solutions, we recommend isolating them in a designated capacity.
Surge protection doesn't guarantee that interactive requests aren't delayed or rejected. As a capacity admin, you need to use the Microsoft Fabric Capacity Metrics app to review data in the throttling charts and then adjust the surge protection background rejection threshold as needed.
Some requests initiated from Fabric UI are billed as background operations or depend on background operations to complete. These requests are rejected when surge protection is active.
Surge protection doesn't stop in progress jobs.
Background rejection threshold isn't an upper limit on 24-hours background percentage. This is because in progress jobs continue to run and report additional usage.
If you pause a capacity when it is in an overloaded state, the System events table in the Microsoft Fabric Capacity Metrics app may show an Active NotOverloaded event after the Suspended event. The capacity is still paused. The NotOverloaded event is generated due to a timing issue during the pause action.
Surge protection doesn't block operations that are billed with Autoscale billing for Spark.