AWS Neptune Benchmark Over Real-World Dataset

Ghislain Auguste Atemezing (1)

1 MONDECA R&D - Mondeca Recherche & Développement

Since the announcement by Amazon of its own graph database service Neptune in November 2017, there has been many expectations on how to compare Neptune with other state-of-the-art enterprise graph databases. Neptune is defined as a high-performance graph database engine supporting popular graph models: RDF and Property Graph Model (PGM). This paper aims at giving an empirical evaluation of Amazon Neptune on real-world RDF dataset from the EU Publications Office (PO). We use three different versions of Neptune (Preview, Neptune 1.0 and Neptune 1.0.1) to evaluate how fast and reliable is the engine with real-world SPARQL queries. We don't resist in comparing some of the results with our previous benchmark with other RDF database graphs, even though one should be careful with such comparison since the hardware settings are not completely equivalent. The results of this evaluation demonstrate that Neptune is faster in loading, perform better in analytical queries, and outperform in updates queries.

Mots clés: AWS Neptune; RDF; SPARQL; Benchmark; Graph Database; Benchmark. Resource type: Benchmark

Domaines: Intelligence artificielle [cs.AI]; Web; Informatique; Base de données [cs.DB]

Ghislain Auguste Atemezing. AWS Neptune Benchmark Over Real-World Dataset. [Technical Report] Mondeca. 2021. ⟨hal-03132794v2⟩