Relational graph data management on the edge: Grouping vertices' neighborhood with Edge-k

As the amount of data represented as graph grows, several frameworks are employing relational databases to manage them. However, the existing solutions store graphs creating a row for each edge in an edge table. In this paper, we propose Edge-k, a novel storage approach that combines additional columns in the edges table, allowing to tune the number of edges stored in a single row by taking into account the overall neighborhood of the vertices, thus providing a better table organization. Compared to the existing approaches in the literature, experiments reveal that our proposal was able to reach a speedup of 66% over a representative real dataset and up to 57% in synthetic datasets when processing Single Source Shortest Path queries. Hence, our solution advances the state of the art in the context of graph data management within relational databases systems.