Assess the performance impact

Reference information for comparing application performance before and after enabling dual ingestion.

Compare the application's behavior before and after you enable dual ingestion. Use the Phase 1 baseline and the success criteria approved for the migration wave.

Dual ingestion produces telemetry for Splunk AppDynamics SaaS and Splunk Observability Cloud at the same time. This increases resource use and network traffic. Validate the change under representative load so that you can tune or roll back before expanding the wave.

Prepare the comparison

Use:

  • The same application scope and environment as the Phase 1 baseline.
  • A comparable traffic profile and observation window.
  • The same metric definitions, aggregation, and percentiles.
  • The application-specific acceptance limits approved in Phase 1.

Compare the measurements

Area Compare
CPU Average, P95, and P99 utilization
Memory Used and maximum memory; garbage-collection frequency where applicable
Response time Average, P50, P95, and P99
Throughput Requests or transactions per second
Reliability Error rate and application restarts
Network Telemetry-related egress volume and rate
Runtime Garbage-collection pauses, thread-pool use, event-loop delay, or another runtime-specific measure
Telemetry pipeline Collector queue growth, dropped data, retries, and export failures

Calculate the relative change for each comparable measurement:

Impact (%) = ((dual-ingestion value - baseline value) / baseline value) × 100

Decide whether to proceed

  1. Compare each result with the approved success criteria.
  2. Confirm that Splunk AppDynamics SaaS monitoring has not regressed.
  3. Investigate changes outside the expected range. Check agent configuration, enabled signals, sampling, resource attributes, Collector queues, retry behavior, and duplicate instrumentation.
  4. Tune the configuration and repeat the test when the application remains stable.
  5. Use the approved rollback procedure if the application breaches a rollback trigger or becomes unstable.
  6. Record the results, changes, exceptions, and approver for the application.
  7. Continue only after the application owner and migration owner accept the result.

Phase 3 performance validation ends when every application in the wave meets its approved limits or has a documented, approved exception.