Software Product System Model: A Customer-Value Oriented, Adaptable, DevOps-Based Product Model.

Abstract

DevOps pipelines have brought notable advantages, such as fast and frequent software delivery to software production paradigms, but dynamically dealing with quality attributes desired by the customer employing a DevOps pipeline remains a challenge. This work aims to define the design of a systems thinking inspired model, called Software Product System Model (SPSM), applying a customer-value oriented, holistic approach for implementing quality requirements, and its application and evaluation in a large software house. The main features include dynamic control of quality gates, the parameters of which are driven by customer requirements and feedback from surveys. All of the inputs are collected in a product backlog and fed forward to the quality gates over the DevOps pipeline. SPSM was successfully deployed in a large software house extending a DevOps pipeline with an accompanying improvement of customer-value oriented key performance indicators for projects. In a 2-year-long case study, security and code quality were the main quality attributes, with the metrics on security vulnerabilities and unit test coverage. At the end of the 2020, the DevOps pipeline within SPSM provided a 69.50% decrease of security vulnerabilities of all software products, and a 29.43% increase in unit test coverage for the whole code base for increasing code quality. At the end of 2020, the project completion ratio was measured to be 99.50% and the Schedule Performance Index (SPI) was measured to be 99.78% as the average of 762 projects delivered. The flexibility of SPSM allowed the software house to adapt to changing customer expectations. A checklist is provided for the replicability of the model application.