Missing Requirements Information and its Impact on Software Architectures: A Case Study

[Context & motivation] In the development of large, software-intensive systems, the system’s requirements are seldom, if ever, concluded upon prior to commencing with systems architecture. Research shows that, in order to manage development and domain complexities, instances of requirements engineering (RE) and systems architecting (SA) processes tend to inter-weave. [Question/problem] However, missing requirements information can cause one to create (or recreate) the needed information during different SA activities. While backtracking in the software development process is known to be costly, the costs associated with missing requirements in the SA process have not been investigated empirically. [Principal ideas/results] We thus conducted a case study where we investigated to what extent requirements or requirements attributes’ information found missing during the SA process and impact of those missing information on SA in terms of effort. The study involved five architecting teams that involve final year undergraduate and graduate students enrolled in the university course on SA, working on architecting a system falls under “banking” domain. Our result shows that, architects did find requirements and requirements attributes’ information missing while architecting. Among requirements information, architects found that, system functionality information, constraints information and system interaction (users/systems) information are missing in requirements at higher percentages. Within requirements’ attributes, architects found requirements priority, dependency and rationale missing at higher percentages. It is also found that, out of total time spent on architecting the system, effort given to recreate missing requirements information is higher for group3 (21.5%), group1 (18%), and group2 (17%) other than group4 (12.37%) and group5(10.18%). [Contribution] The anticipated benefits of the findings are, it can motivate researchers to venture into other areas of software engineering (such as coding, testing, maintenance, etc.) from the view point of missing requirements information and its impact on those areas. This knowledge could help software practitioners to decide what kind of information need to take care of, during RE process, that could possibly ease SA process and later development phases. To the best of my knowledge, this is the first work which focuses on, to what extent requirements and requirements’ attributes information found missing during SA; characteristics and impact of those requirements missing information on SA process in terms of effort