I did some experiments on four graph database systems: neo4j, DEX, infinitgraph, orientDB. I think it might help you. I used very small graph for evaluation when compared to yours. You can find the results from my experiment in 7th slide. I ran pagerank for ranking nodes and SCAN for clustering nodes. Don't mind if something is wrong in it.
Venkat On Wed, Jun 29, 2011 at 12:54 PM, Aliabbas Petiwala <[email protected]>wrote: > greetings neo4j folks! > > we would really like to go for neo4j for our social networking website but > we are still not convinced about the scalability and concurrency of neo4j > considering that it may have to deal with billions of peta bytes of data. > a > massive write load on master is predicted due to millions of users updating > their status \ other information. > > infinitegraph.com claims high scalability and distributed servers whereas > neo4j constrains us on a single database and the master seems to be the > bottleneck in writes. > > have any tests or simulations being performed on neo4j to gauge its > efficiency and fault tolerance in large scale systems on the scale of > facebook and twitter? > > orientdb is also promising since its open source but has only native > support > for graphs and has no spatial features which we require. > > hope neo4j folks convince us before we take a major decision of choosing > the > right graph db . we are also open to polyglot persistence involving a > combination of neo4j, orientdb and cassandra is it wise to go for such a > combination ? > > -- > Aliabbas Petiwala > M.Tech CSE > _______________________________________________ > Neo4j mailing list > [email protected] > https://lists.neo4j.org/mailman/listinfo/user > -- Thanking you, Sincerely, M.VenkataSwamy, Graduate Research Assistant, UALR
_______________________________________________ Neo4j mailing list [email protected] https://lists.neo4j.org/mailman/listinfo/user
