Optimal cost: a criterion to release multiple versions of software

With successful multi up gradations of software as an indispensable part of any Information Technology company, the challenge before them is to decide when to stop testing these multiple versions. Too much of testing delays the software system’s entry into the market and also costs more. Releasing the software system too soon without adequate testing may cause customers’ dissatisfaction and may lead to failure of the software system in the market. In this paper we introduce cost model for each release of software based on debugging costs at different phases of software development viz-a-viz, testing and operational phase. Our model incorporates three new features that are not included in the literature. First, to the total cost incurred during the testing phase of new release, we add the debugging cost associated with remaining number of reported faults from previous release. Second, we consider that some of the remaining faults of previous release are partly removed during the testing phase and partly during the operational phase of new release. Third, we calculate the optimal testing time for each release of software in the presence of debugging costs associated from the previous releases. An Algorithm for finding optimal testing time for each version is developed. We estimate the parameters of the model using Statistical Package for Social Sciences on real data set and obtain optimum stopping time for each version of software using Maple software.