Integration of a Software Asset Management Tools into Heterogeneous Environments

Data warehouses employ Extract-Transform-Load (ETL) processes to integrate information from heterogeneous sources into a single repository, thus supporting management decisions with a global view on data across the organization. Deriving, documenting and validating requirements is referred to as requirements engineering (RE). The research community emphasizes the importance of thorough and well-executed RE in data warehouse projects. While existing research focuses on theory, little research has been conducted on applying RE methods in the context of ETL. This thesis studies the design and implementation of a RE process for the ETL component of a data warehouse application specialized in Software Asset Management (SAM). Based on Sommerville’s and Kotonya’s linear RE process and other methodology suggested in literature, a 4-phase RE process was designed: First, a planning phase identifies problem statement, system boundary, actors and stakeholders, but also sources and elicitation techniques for requirements. Second, the chosen elicitation techniques are employed and collected information is grouped into a list of functional and non-functional requirements. Third, analysis workshops with technical stakeholders facilitate checking requirements for necessity, consistency, completeness and feasibility. Lastly, final requirements are validated, prioritized and signed off during a workshop with product managers. This process is implemented and evaluated. The results suggest that adding a dedicated planning phase to Sommerville’s and Kotonya’s process helps to ensure that only relevant requirements are collected and all available sources are being taken into consideration. Involving technical stakeholders was crucial to getting the requirements “right”. Precisely documenting planning results, collected information and requirements in parallel to the RE process increased effort but had an positive impact on requirements’ quality while enabling traceability of requirements back to their source.

Project information



Thesis for degree:



Nils Christian Hans