Application Performance Management the Case for Adaptive Instrumentation in J2EE Environments
Source: Precise Software Solutions
Measuring and isolating performance bottlenecks in production J2EE applications requires an efficient data collection technology that provides detailed performance metrics with minimal overhead. Excessive overhead is usually the result of over-monitoring; therefore, deciding which components of the application should be monitored is a key issue. Traditionally, this has been a manual trial-and-error process, where the application structure is examined and components that could potentially cause a bottleneck are added to the list of monitored components. If undetermined performances issues still exist after all the components on the list are monitored, a manual analysis process is employed until the root cause of the remaining performance problems is found. During this process, if too many components are monitored, the associated overhead will affect the system.