User Acceptance Testing
I am working on a COTS iterative implementation project. It has been decided that User Acceptance Test will not be included as a separate test stage but rather as another component of all previous stages of testing.
If anybody has experience with or knowledge about this approach, please share your experience or point me to resources on how to go about implementing it.
Re: User Acceptance Testing
First, I'm guessing you might get more opinions in the Functional Testing forum.
1) Is the COTS going to require modification and/or custom software for implementation?
2) If yes - are the changes phased in?
3) Is this project strictly integration of COTS without customization?
4) Has the COTS been accepted as-delivered?
1) I don't believe your first statement makes much of a difference in the approach.
2) The approach varies according to your answers to the above questions.
3) A general approach that may get you started:
3.1) Determine your high-level test requirements as driven by what must work along each stage or phase of COTS integration.
3.2) This sets you up to be able to synchronize test sets with the schedule of implemented piece/parts, and - allows you to know what tests need to be developed to detail and in what order. It also allows you to identify the resources you need for readiness at each implementation milestone.
3.3) At the completion of your testing for each implementation milestone - you can (assuming all tests pass) have each component signed off as acceptable**. This is often known as Preliminary Qualification Testing or PQT. Repeat this for each implementation layer. For each layer, repeat all tests for all layers implemented up to this point. When all parts of the implementation are completed - you could possibly get acceptance of the whole.
4) When PQT is signed off - move onto FQT (Formal Qualification Testing). This is usually just re-proving (repeat of PQT or a subset of key PQT tests) what is already known in a formal setting where tests may need to be witnessed and certified.
5) Following FQT, there may be an Operational Evaluation which is typically a test of the as-delivered systems, applications, subsystems, etc. - in the actual intended environment. One could say this is the acceptance test. Ths test is proof that the system operates as intended and, that any known and accepted issues have acceptable workarounds.
** Acceptable only for PQT. FQT should follow.
Let us know if this helps.