Graph database that supports efficient sorting and/or range retrieval

Using Twitter's data model a开发者_如何学编程s an example:

User A follows N other users, each of whom posted N different posts. Posts are timestamped.

Is there a data store or graph database that supports efficiently reading the latest N posts from users that User A is following?

It seems best that posts are stored in sort order (like a Cassandra ColumnFamily or a Redis ZSET).

Also, the ability to shard the data across different machines is critical.

Have a look at Neo4J, it should fit your needs perfectly. One recommendation though, use its REST interface since it gives you the freedom to chose your weapon uhm programming language and abstracts a lot of the complexity of working with the Java bindings directly

An option for you might be ... RDF Databases (AKA triple stores/quad stores) and try SPARQL to query graphs and retrieve subgraphs from your data. See also SPARQL by example.

One that scales up to billions of triples or quads is 4store, it is written in C and allows data sharding across segments, It makes the most of commodity clusters or multi-core architectures. There are a wide range of client libraries to interact with it from Java, perl, Python, ...