Configure MetricSets
Reference information for configuring and verifying MetricSets for Splunk APM.
Configure MetricSets for span attributes required for APM troubleshooting, dashboards, detectors, or alerting.
Splunk APM derives request, error, and duration metrics from traces. Splunk APM does not expose every span attribute as a metric dimension by default. Index only the attributes required by the migration wave because each additional dimension can increase metric cardinality.
Prerequisites
Confirm that:
- Representative traces contain the attributes you plan to index.
- Dashboard and detector requirements identify the necessary filters and groupings.
- The attributes you plan to index have stable, bounded values appropriate for metrics.
- An APM administrator can manage MetricSets.
- Review applicable entitlement and retention requirements.
- During MetricSet configuration, run the cardinality analysis for each proposed MetricSet. Activate the MetricSet only when it shows the Within Entitlement status.
Select the MetricSet type
| Requirement | MetricSet |
|---|---|
| Use an indexed span attribute in Tag Spotlight or live troubleshooting | Troubleshooting MetricSet |
| Use a dimension in dashboards, detectors, or long-running monitoring | Monitoring MetricSet |
| Use the attribute for both troubleshooting and monitoring | Create the applicable Troubleshooting and Monitoring MetricSets |
See Learn about Monitoring MetricSets and Use and manage Troubleshooting MetricSets.
Determine the required dimensions
Use the migration inventory and target queries to identify dimensions such as:
| Splunk AppDynamics scope | Candidate target dimension |
|---|---|
| Application or business domain | service.namespace or the approved environment dimension |
| Tier | service.name |
| Node or service instance | service.instance.id or another approved instance dimension |
| Business Transaction | sf_workflow |
Confirm current default dimensions before creating a custom MetricSet. Do not index a dimension solely because it existed in Splunk AppDynamics.
Configure and verify MetricSets
- Open a representative trace and confirm that the required attribute and expected values appear.
- Estimate the number and growth of distinct values.
- Reject attributes containing request IDs, user IDs, raw URLs, timestamps, or other unbounded values.
- In Splunk Observability Cloud, select .
- Create the required Monitoring or Troubleshooting MetricSet with the approved service scope.
- Wait until the configuration becomes active and new telemetry reaches the MetricSet.
- Confirm that the dimension appears in the intended dashboard, detector, or troubleshooting workflow.
- Verify that cardinality and entitlement consumption remain acceptable.
- Record the attribute, MetricSet type, scope, owner, purpose, and validation result.
MetricSets do not retroactively index historical traces. Validate with telemetry received after the configuration becomes active.
This task is complete when every required target filter or grouping uses an available, validated dimension and you avoid unnecessary high-cardinality indexing.