Test cases granularity

Question

How much granularity and importance could be used in test cases. Either in unit tests and integrated tests.
Particularly about negative cases. There are possibility that we could think of a lot of negative cases for testing a simple function. How could this be controlled?
Also at what level should the testing be done considering a 3 layer architecture?

Answer 1

This question needs a complex answer. The granularity depends directly from the budget allocated for testing. I can suggest that category partion can help you to find relevant test cases.