Framework for Testing Strategy in Software Product Development
Volume 17, Number 3 Article by Ashish Tewary and Janat Shah September, 2005
Framework for Testing Strategy in Software Product Development :
Traditionally in software development, testing has been seen as a mechanism to detect faults or deviations from the design specification. But this narrow focus does not address the business issues faced by software product companies at large. But recent studies reveal that testing has been used by leading software product companies as a strategy to sense customer needs, test alternative technical solutions, and integrate the knowledge gained from markets and technologies.
Three types of testing are generally performed in a typical software product development: unit testing, that involves the basic testing of a piece of code of a component; integration testing, to check whether the modules integrated through interfaces are working consistently to implement a pre-decided set of functionality, and beta testing that focuses on validation by detecting whether the development team is building the product of desired specification. The chosen testing strategy must be designed to reflect the realities of the environment which the product has to deal with. The most significant parameters affecting the choice of testing strategy are fidelity of tests conditions, cost of testing, cost of rework, cost of delay and cost of software quality. These parameters differ across types of testing. Despite the fact that the advantages of early problem identification and solving through the testing of prototypes have been established, research in this area has not looked at testing in a comprehensive manner.
Ashish Tewary and Janat Shah develop a framework to derive the optimal testing strategy for different types of testing. The framework helps in probing the pros and cons of different testing strategies, respecting the complexity of different types of testing, and provides guidance for making the final choice. The framework, the authors believe, will result in effective managerial decision-making, while adding some new building blocks to this growing area of academic research.
Reprint No 05307c