Process Tailoring in Large-Scale Agile Programs

Background: Teams in large-scale agile programs need to achieve collaborative software development. A proposed guide to collaboration is effective coordination. Large-scale agile software development is well accepted in the software industry, but there is little understanding of such projects and programs achieve effective coordination in autonomous cross-functional teams. Therefore, I conducted a case study of a large-scale software program consisting of seven autonomous teams in an organization. Five of them was DevOps teams, where DevOps is merely a team composed of developers who are working on the development and operational tasks. Aim: The thesis aims to investigate what dependencies and their related agile practices that act as coordination mechanisms to facilitate the large-scale agile development. Additionally, the aim is also to recommend a starter set for providing coordination in the large-scale by using a dependency taxonomy, which mapping agile practices. Method: A qualitative case study was conducted for the research design. The data was collected by conduction 40 observed meetings, as well as 18 entire working days in the organization's open work area to observe them in their everyday work. A dependency taxonomy was used to map and categorize the coordination mechanisms. Results: The results revealed that there were 34 coordination mechanisms and 77 pairs of dependencies presented in the program. The coordination mechanisms could be mapped into three categories, with subcategories each: knowledge dependency, process dependency, and resource dependency. The knowledge dependency was predominant with the frequency of 73 % of the three categories. These means that focusing on selecting agile practices that address the types of knowledge dependency should be recommended for providing coordination. Furthermore, the results revealed that 12 agile practices would be a good choice for coordinating and tailoring a large-scale program. Conclusion: It is possible to use a dependency taxonomy to map coordination mechanisms in a large-scale agile program. The coordination mechanisms made collaboration between the teams in the program, by implement Scrum of Scrum meetings, daily stand-up meetings, demo meetings, Sprint Planning meetings, and introduce different roles, such as project managers, team leads, Product Owners and DevOps developers. These mechanisms lead to fast Sprint periods, frequent production setting, a common understanding of what is being created, and autonomous decisions in the program