Background: DevOps is a software development methodology that is adopted by many firms to achieve increased organizational performance and maintain competitive advantage. There are many academic publications on this topic, but no consensus on the definition of the methodology and the practices that constitute it. The research primarily focuses on the organizational roles of developers and operations, ignoring other roles which may be relevant to the successful adoption of DevOps and the realization of its benefits.

Objectives: The goal of this research is to create a comprehensive model of DevOps methodology and the practices that constitute it, to define the organizational roles that need to be involved in DevOps implementation, and to explore the relationship between the inclusion of the additional roles and practices and the organizational performance.

Methodology: We employed a multi-method research approach, adopting both qualitative and quantitative research methods. The qualitative research consists of a Systematic Literature Review (SLR), conducted in the form of a Tertiary Study, and a series of semi-structured interviews, conducted with DevOps industry experts. The qualitative research produced a conceptual model, describing DevOps practices and capabilities, as well as their relationship with the roles that perform them, which then was subjected to quantitative research. The quantitative research used HELENA survey data set as a secondary source of large-scale data on the usage of practices and organizational performance to investigate the connection of their usage with organizational performance and goal achievement.

Results: The qualitative part of the research has resulted in a conceptual model of DevOps, consisting of 175 practices spanning 5 DevOps CALMS dimensions and 32 characteristics. For many of the practices, even the involved organizational roles were established, resulting in 21 roles. Besides developers and operations, the roles of executives and management were prominently represented in the model, with their involvement focused on creating the right organizational culture. The qualitative phase has confirmed the positive association of the usage of DevOps practices with organizational performance, but there was insufficient data to explore performance associated with the added DevOps practices or roles.

Conclusions: This research contributes a comprehensive list of DevOps practices and fills a significant gap in the literature with regard to the research on organizational roles in DevOps. Firms on their DevOps adoption journey should consider the suggested practices and the broader set of roles to fully realize the benefits of the methodology.

Recommendations for future research: We recommend further qualitative and large-scale quantitative research of the proposed DevOps model, with a focus on the additional practices not covered in the academic literature, the involvement of the broader organizational roles, especially executives and management, and the relationship between the adoption of the additional practices and involvement of the additional roles and the achievement of organizational goals.