Lightweight Transaction Profiling represents the best value in terms of minimal monitoring overhead with excellent visibility into the performance of your application and its individual transactions. However, there are times where additional diagnostics might be required:
  • Your application is distributed, so you want to trace transactions across JVMs.
  • Profiling points to a problem that requires additional context to diagnose; for example, profiling points to an SQL query as the root-cause of slowness, so you want to capture the specific SQL being executed.
  • Profiling shows a wide discrepancy between max and avg times, so you want to trace individual transactions and identify why some are poor performing in comparison to other that are not.

loading table of contents...