Scaling DBSCAN towards exascale computing for clustering of big data sets