Distributed Graph Layout with Spark

Dessin de graphe distribué avec la bibliothèque Spark

Antoine Hinge (1) , David Auber (1)

1 LaBRI - Laboratoire Bordelais de Recherche en Informatique

This paper presents a novel way to draw very large graphs, especially those too big to fit the memory of a single computer. This new method takes advantage of the recent progress in distributed computing, notably using the Apache MapReduce library called Spark. Our implementation of a force-directed graph drawing algorithm and the way to compute repulsive forces in MapReduce are exhibited. We demonstrate the horizontal scalability of this algorithm and show layouts computed on a Hadoop cluster with our method.

Mots clés: Graph Drawing; MapReduce; Hadoop; GraphX; Spark; Distributed Computing

Domaines: Calcul parallèle, distribué et partagé [cs.DC]; Autre [cs.OH]; Réseaux sociaux et d'information [cs.SI]; Géométrie algorithmique [cs.CG]; Algorithme et structure de données [cs.DS]

Antoine Hinge, David Auber. Distributed Graph Layout with Spark. Information Visualisation (iV), 2015 19th International Conference on, Jul 2015, Barcelone, Spain. pp.271--276. ⟨hal-01187421⟩