Performance testing is the structured technical process of measuring a system’s speed, throughput, and stability under defined load conditions — validating that the system meets its non-functional performance requirements before it is used in a live operational environment. In ERP and EPM implementations, performance testing is not a single test run; it is a phased test programme covering interactive user performance (how fast do screens load, forms save, and reports render for a defined number of concurrent users) and batch process performance (how long do data loads, calculation rules, and consolidation runs take at full production data volumes).
Performance Test Types
| Test Type | Purpose | Input |
|---|---|---|
| Load test | Validate system performance at expected peak concurrent user load | Peak concurrent user count from NFRs; realistic user journey scripts |
| Volume test | Validate batch process performance at production data volumes | Production-equivalent data set; batch process inventory from NFRs |
| Stress test | Identify the point at which system performance degrades unacceptably | Load incremented beyond expected peak until failure threshold is found |
| Endurance test | Validate system stability over an extended period — typically 8 to 24 hours of sustained load | Sustained load at expected average; memory leak and resource degradation monitoring |
For Oracle EPM Cloud applications, the most critical batch performance tests are the calculation rule execution time (for large, complex PBCS or FCCS applications), the consolidation run time (for FCCS with many entities and complex elimination rules), and the data load duration (for FDMEE or Data Management loads from large ERP extracts). These are the processes that directly determine whether the close cycle can be completed within the required window — and they must be tested at production data volumes, not at the smaller datasets used during system integration testing.
Common Gaps and Failure Modes
The specific failure that produces performance problems discovered post-go-live is performance testing run against test data volumes rather than production-equivalent data volumes. An Oracle FCCS consolidation run that takes 12 minutes against 20 test entities and 6 months of data will take materially longer against 60 live entities and 24 months of historical actuals, with full elimination rule complexity. The test passes. The live consolidation takes 4 hours. The finance team cannot complete the close on time. Performance testing at sub-production data volumes does not test performance; it tests functional correctness at low data volumes. These are different things and should not be conflated.
How Loop Wise Solutions Produces This
Loop Wise Solutions designs the performance test plan from the NFR document — mapping each measurable NFR threshold to a specific test type, scenario, and acceptance criterion. We insist on production-equivalent data volumes for all batch performance tests, and we schedule performance testing early enough in the programme that identified performance issues can be remediated through configuration optimisation — restructuring calculation rules, adjusting consolidation perimeter design, or scaling infrastructure — before the go-live date is at risk. Performance issues found in performance testing are fixable. Performance issues found in production are operational crises.
