Although it could appear like a counterintuitive concept, the idea of performing software testing in production has gained more and more visibility in a software development world that aims for rapid delivery of new features and where it could be more and more difficult to reproduce the full complexity of applications in a separate environment. In this article, Marc van ’t Veer discusses the concept of testing in production and why it should be performed.
Testing in production can be defined as testing on a real live system, either about to go live or with live users. The article explains however that testing in production is not a replacement other global tests such as system or acceptance tests. It should not be done because there was no more time left or because it is cheaper to do than previous test stage.
The article is based on the development of an API project performed at a large Dutch mobile telecoms provider. This application met many problems going live as the stability of the test environment didn’t meet the instability of the production context. Marc van ’t Veer explains that the fact of testing in production is not a new concept and that it has been applied by consumer organisations around the world for decades.
The conclusion of the aricle is that “When TiP is introduced into an organization, the whole development model changes. Before TiP is added then it is mainly BUFT or “big up-front testing” that is executed. This means that all tests are done before the software goes live (BUFT). But, with TiP, the developers have more direct contact with the end users and the testers do not need to act like end users. ”
Read the complete article on https://www.polteq.com/wp-content/uploads/2013/02/testingexperience20_12_12_Marc_van_t_Veer.pdf