ROLAP and DOLAP Systems: A Comparative Study

Abstract

NoSQL databases have been gradually adopted driven by the growth of Big Data. However, its application to Business Intelligence suggests a transfer of concepts typically associated to the relational database world. To build data warehouses in Business Intelligence it is employed the dimensional model. One of its goals is to facilitate the interactive exploration of data. The adaptation of the dimensional model to NoSQL databases is relevant given its proliferation and the data navigation advantages that the dimensional model provides. It is therefore urgent to make more studies that allow assessing this feasibility. In this work, a dimensional model is extrapolated to a model in a document data warehouse. The rising questions are: (i) using document databases what is the feasibility of models that simulate the dimensional model; (ii) what the specificities in data preparation phase are; (iii) what the degree of simplicity in writing queries for data analysis is and (iv) what about its performance. In this context, the same set of analysis questions are applied to two systems: one called ROLAP (relational) and the other DOLAP (document). The results achieved reveal that data analysis queries for document warehouses are not complex, although with different execution times. On the other hand, depending on the adoption of the type of dimensional model, there are different efforts in the data preparation phase.