Architectural Evolution of Stream Data Warehouses: A Systematic Literature Review and Research Perspectives

Stream Data Warehouses (SDWs) were introduced to integrate traditional data warehousing with continuous processing of incoming data streams. Such capabilities are increasingly relevant to computational intelligence systems operating in dynamic, data-intensive environments, where continuous data analysis, high performance, and fault tolerance are essential. This study investigates the evolution of SDW architectures and their relevance in the era of cloud-based data platforms. A systematic literature review of Scopus, IEEE Xplore, and ScienceDirect identified 102 unique publications, from which five complete SDW architectural models were selected for detailed analysis and compared with Data Lakes, Data Lakehouses, and Modern Data Warehouses. The results reveal a shift from specialized SDW models towards flexible architectures composed of cloud and data-processing services. While cloud platforms are generally considered scalable, available, and technologically flexible, stream-specific capabilities such as on-the-fly processing, continuous querying, low-latency analytics, and integration of current and historical data are often delegated to additional technologies. This may increase technological dependencies and integration complexity. The findings suggest that advancing analytical infrastructures for computational intelligence requires combining the benefits of cloud platforms with further development of dedicated architectural mechanisms for stream data warehouses, rather than relying exclusively on the composition of existing services.