An approach to integration testing

A while ago I answered the following question on’s Ask The Software Quality Expert: Questions & Answers.

I'm working as an IT engineer and I have a software integration testing question. I have very low-level requirements which match directly to a function or some statements of function -- that is, a group of requirements matches to one function. I want to do software integration testing for the above requirements. Can you please suggest to me how integration testing is possible? I want to know the testing approach.

Here is a clip from my answer:

When talking about your question with Dale Emery, Dale pointed out that for him, one of the fundamental ideas for integration testing is the idea of disagreement between various parts of the system. That is, if we believe part X does its specific function well (hopefully based on some level of unit testing) and part Y does its specific function well, then when we put them together and test we're looking for ways in which there is conflict or unintended consequences between the two. The less certain we are about the different parts we'll be testing, or the lower our confidence in how those two parts will interact, the more integration testing we might want to do.

You can find the full posting here .