On the impact of data distribution in federated SPARQL queries

With the growing number of publicly availableSPARQL endpoints, federated queries become more and moreattractive and feasible. Compared to queries against a singleendpoint, queries that range over a number of endpoints posenew challenges, ranging from the type and number of datasetsinvolved to the data distribution across the datasets. Existingresearch focuses on the data distribution in a central storeand is mainly concerned with adopting well-known, traditionaldatabase techniques. In this work we investigate the impact of thedata distribution in the context of federated SPARQL queries.We perform a number of experiments with four federationframeworks (Sesame Alibaba, Splendid, FedX, and Darq) againstan RDF dataset, Dailymed, that we partition by graph and class.Our preliminary results confirm the intuition that the moredatasets involved in query processing, the worse performanceof federation query is and that the data distribution significantlyinfluences the performance.