Speaker
Description
Controlling CERN’s accelerator complex requires a significant number of domain specific applications. Typically these are developed by small teams comprised of a staff member with essential domain knowledge, and 1 or 2 students or graduates staying for 1-to-3 years. Developments may span over months or even years, according to the complexity and requirements, which may evolve significantly over time to follow the accelerator programme. After some years, applications need to be consolidated or re-written to follow the inevitable technology evolution and remain secure, hardware compatible, and based on technologies for which skills exist in the marketplace. This development approach has produced results, but is inefficient due to the isolated developments sometimes leading to black-box solutions that eventually need reverse-engineering. Furthermore, long-term staff members can become siloed within specific domains, increasing the risk of single points of failure. This paper shares the results and experience gained over 18 months in applying a more agile, task-force team-based approach to address the aforementioned limitations and inefficiencies. Together with the application of micro-frontend architectures, the team has been able to develop and maintain multiple applications with enhanced distribution of both domain and technical knowledge, as well as predictable delivery times. The team’s workflows will be described, together with challenges faced and lessons learned.
