The approach of solving performance issues is to try to characterize the regression first by investigating the situation by following a specific process. After the constraining resource has been identified, further steps to resolve the situation are applied. After potentially fixing the situation, the measurement in production needs to be repeated. It's important to not change behavior or configuration without the verification that the changes in fact provide the expected results.

The jPDM approach investigates performance regression without considering the application's code, by applying a uniform solving process.

What tools and metrics are needed to apply this approach?

Depending on the system in production, tools that ship with the operating system, as well as Java Runtime-related tools, are useful. Since all aspects consider the overall system at the operating system level rather than just the application alone, operating system tools and lower-level metrics are more helpful than application-specific ones.

However, the technical metrics of the application, such as response time or throughput, are the first place of focus that indicate the application's quality of service. If these metrics indicate a performance issue, then it makes sense to investigate using lower-level metrics and tools.

The next section examines how to gather the application's technical metrics.