A reliable service behaves predictably even when things go wrong: incomplete input, slow or failing dependencies, unexpected external responses. Reliability in Designer comes from validating early, handling errors intentionally, treating external systems defensively, and proving all of it with tests.
This page covers practical best practices for multi-level testing. Testing is most effective when it covers happy paths, edge cases, failures, and integration behavior with repeatable scenarios tied to business expectations.
Test Across Multiple Levels
Service quality improves when testing is built into development instead of deferred until the end.
-
Model-level testing: Verify paths, decisions, mappings, and expected outputs. Implement unit tests with TestableClass; the same tests double as regression tests on every subsequent change.
-
Business scenario testing: Test realistic end-to-end use cases, including alternate and exception flows.
-
Integration testing: Verify behavior with connected systems, adapters, and APIs.
-
Regression testing: Re-check existing behavior after every significant change. You can use the Scheer PAS Analyzer to do this.
-
Negative testing: Confirm the service fails safely with invalid, missing, or conflicting input.
-
Final acceptance testing: A last business sign-off round shortly before go-live, run against realistic data and scenarios.
Minimum expectation for robust services: Every critical path, every important business rule, and every external dependency should be covered by repeatable test cases.
Related Content
Related Documentation: