Monitoring and controlling release readiness by learning across projects

Abstract

Releasing software on time, with desired quality while staying within budget is crucial for success. Therefore, product managers should proactively know which release readiness attributes are not performing sufficiently well (i.e., bottleneck factors) throughout the development cycle and consequently may limit readiness of the software release. We present the Cross-project Analysis for Selection of Release Readiness attributes (CASRR) method to help project managers in (i) systematically studying and analyzing release readiness attributes across multiple projects, (ii) selection of release readiness attributes for monitoring which have previously been shown to become bottlenecks in similar projects in the past, and (iii) learning how bottleneck occurrences are influenced by project characteristics. We applied CASRR to two Open Source Software projects, and analyzed six release readiness attributes in 34 similar projects over a period of two years. Continuous integration rate, feature completion rate, and bug fixing rate are observed as the most frequent bottleneck factors. Bottleneck occurrences of the monitored release readiness attributes are significantly influenced by the maturity of a release. Furthermore, the continuous integration rate is found to be significantly influenced by the team size.