Performance Levels

When a load test is completed, a performance engineer usually has to assess whether the test has been successful or not. This assessment is typically based on a thorough analysis of the load test results.

Performance engineers can use the broad range of analysis tools that the Workbench and Performance Explorer provide. They can use the various reports to analyze a number of metrics, such as the amount of errors or the response time thresholds. They could also compare response times to performance criteria or service levels, which are usually determined prior to the load test execution.

This process can be described as manual load test assessment. However, in many cases an automatic assessment of the load test is useful. For example: In continuous integration environments, users typically want to be notified only in case performance criteria have been violated. In Silk Performer, you can use the so-called performance levels for such scenarios. Performance levels can be defined in two ways: proactively and reactively.

Of course, these approaches can also be combined. This might be necessary, if performance levels have been defined prior to a load test execution and need to be adapted when the execution is completed, because the reports do not reflect the desired status.

To cover both approaches, you can access the performance level dialogs through the Workbench and Performance Explorer. In the Workbench, click Project > Define Performance Levels or Project > Assign Performance Levels. In Performance Explorer, click the Reports tab and click Define Levels or Assign Levels.