> > Is there any recommendation on when to use additional node/relationship > > connections versus additional node properties? > > It depends of use case but go with relationships if it "enriches" the > graph with information that makes sense. >
And this related to another aspect, based on the balance between two factors: - Total data size (the graph approach increases the primitives count which can be an issue for scalability and performance) - Need to perform graph-like operations on the data I'm also thinking about exactly this issue because it is related to the idea of storing map data in the graph. Most databases store geographic objects as blobs that do not benefit from the database itself, and include internal opaque facilities for operating on that data. But with neo4j the option exists to store this as a sub-graph, exposing the structure and operations to the developer (for better and worse :-). See 'Neo4j Spatial<http://wiki.neo4j.org/content/Neo4j_Spatial>' for some initial discussions on this. So, in my opinion the fact that you said the data would be written infrequently seems to imply that you will not be making much use of the graph nature of the data, and then you might as well store it as properties and benefit from fewer primitives. I see Johan also suggested a blob / array. I guess it could even be a byte[] from serializing the entire hash. _______________________________________________ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user