MASTERÂ Assignment
Towards a Data Fabric Architecture Reference Capability Model
Type : Master M-BIT
Period: January - July, 2026
Student: Oostrik, S.H. (Stan, Student M-BIT)
Date Final project: July 2, 2026
Supervisors:
Abstract:
Data fabric has emerged as an architectural paradigm for enabling governed access to distributed data across complex enterprise landscapes. However, because data fabric is still an emerging concept, there is limited consensus on what capabilities constitute a data fabric, how it differs from related architectures such as data lakehouse and data mesh, and when it should be used. This paper addresses these gaps by developing two design artifacts: a Data Fabric Architecture Reference Capability Model (DFA-RCM) and color-coded federation variants for comparing data fabric, data lakehouse, and data mesh. Following design science research, the artifacts were developed through a problem investigation with enterprise architects, a literature review on data fabric architecture, and three iterative design cycles based on Van Riel et al.'s method for developing generic capability maps. The DFA-RCM was modeled in ArchiMate and identifies core data fabric capabilities, including Data Policy Management, Metadata Acquisition, Metadata Management, Metadata Access, Connectivity Management, Virtual Data Integration, and Data Access. The federation variants compare the three architectures by classifying shared Level 1 capabilities as centralized, hybrid, or decentralized based on coordination and execution. The evaluation indicates that most design requirements were realized within the scope of this research. The DFA-RCM supports target capability modeling, while the federation variants support architectural discussions about when data fabric is appropriate. However, the variants do not yet provide a complete decision framework. The findings suggest that data fabric occupies a hybrid position between centralized and decentralized architectures by combining central governance, metadata, and access capabilities with distributed storage and execution.