Real-time data warehousing for business intelligence

Abstract

Real-time ETL and data warehouse multidimensional modeling (DMM) of business operational data has become an important research issue in the area of real-time data warehousing (RTDW). In this study, some of the recently proposed real-time ETL technologies from the perspectives of data volumes, frequency, latency, and mode have been discussed. In addition, we highlight several advantages of using semi-structured DMM (i.e. XML) in RTDW instead of traditional structured DMM (i.e., relational). We compare the two DMMs on the basis of four characteristics: heterogeneous data integration, types of measures supported, aggregate query processing, and incremental maintenance. We implemented the RTDW framework for an example telecommunication organization. Our experimental analysis shows that if the delay comes from the incremental maintenance of DMM, no ETL technology (full-reloading or incremental-loading) can help in real-time business intelligence.