Performances of OLAP Operations in Graph and Relational Databases

The increasing volume of data created and exchanged in distributed architectures has made databases a critical asset to ensure availability and reliability of business operations. For this reason, a new family of databases, called NoSQL, has been proposed. To better understand the impact this evolution can have on organizations it is useful to focus on the notion of Online Analytical Processing (OLAP). This approach identifies techniques to interactively analyze multidimensional data from multiple perspectives and is today essential for supporting Business Intelligence. The objective of this paper is to benchmark OLAP queries on relational and graph databases containing the same sample of data. In particular, the relational model has been implemented by using MySQL while the graph model has been realized thanks to the Neo4j graph database. Our results, confirm previous experiments that registered better performances for graph databases when re-aggregation of data is required.