Leveraging data monitoring to aid decision-making in large scale continuous integration systems

• Autor(a) principal: Terrantez, Diogo António Ferreira
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.22/21415
• topic: Continuous Integration; Metrics; KPIs; Data Monitoring

Following the Android Open Source Project (AOSP) initiative, Bayerische Motoren Werke GmbH (BMW) has started developing the head units for their latest generation vehicles with an Android-based operating system. While providing multiple benefits, it brought about a new development methodology much different than developing for AUTOSAR, the industry standard and previous methodology. To accommodate this new methodology, a new and highly complex Continuous Integration (CI) infrastructure was designed from the ground up. Managing such an infrastructure has proved arduous and existing monitoring solutions failed to meet the project’s demands. As such, a need has emerged for a new, more complete monitoring solution, that can accurately measure and provide insight into the state of this infrastructure. This work aims to provide such monitoring system that, by aggregating relevant metrics, is able to formulate meaningful KPIs that may be leveraged during the decisionmaking process for infrastructure maintenance and future project growth. The research section for this work consists in the investigation on how to best define metrics and KPIs, and relevant monitoring tools and other technologies, that may be used to deploy and configure the required monitoring solution. Besides this investigation, A literature review on relevant topics and collaborator interviews were conducted in order to identify problems with the CI infrastructure and metrics that should be measured as part of the solution. A proof-of-concept was developed as a way to verify the solution’s viability, resulting in the implementation of some metrics and definition of KPIs. A possible AWS (Amazon Web Services) deployment was also entertained. Results from the implementation suggest that the problem can be solved trough the implemented solution, though some limitations afflicted the development workflow. These are, however, already being addressed and should not pose a problem in the future. The main contributions from the work revolve around features implemented for the monitoring solution, which are already improving daily workflow for a number of collaborators.