A hybrid OLAP & OLTP Architecture Using Non-Relational Data Components

Abstract

Relational database systems are still the first choice for the realization of business application systems. They are used as operational databases to persist business process data or as the basis for data warehouse systems. In recent years, the NoSQL-movement proposed a number of new non-relational data stores because of changing requirements of Web 2.0 applications and related limitations of relational systems. We argue that these new systems can also be used in OLTP- and right-time OLAP-scenarios. We demonstrate how enterprise application systems can be realized based on concepts of the NoSQL-movement. We propose an architecture that enables the deployment of an appropriate data model to a certain business domain and the integration of OLTP- and OLAP-functionality. Components of the architecture are suited to enhance business data with analysisrelevant information and to perform analysis tasks. To demonstrate the application of the architecture and its analytic features, we introduce a short case study.