Pyramid, Honeycomb, Feather Testing Models
How much time should be allotted to each level of test activity (unit test, integration test, system test, end-to-end test)? it varies; but definitely just enough to get the job done to an agreed level of Acceptance Credence. A model that is not adaptable to the task at hand constricts development, and needlessly constricts the actors themselves(coder, tester) -- such that, either (a) the team starts to get coerced to follow the model, sacrificing any innate efficiency; or (b) the team ignores the model and lets the nature of the task dictate the most efficient work flow. The nature of the test_paths themselves would provide the impetus to place emphasis on either (c) integration tests (e.g. majority of the test_paths are integration subsystems); (d) a balance of unit tests and end-to-end tests (e.g. the product is simple, or a self-contained supersystem); or (e) a balance of unit, integration, and comprehensive systemic tests ...