AWS Neptune is a graph database service offered by AWS.
Performance
- Neptune query performance is primarily driven by complexity rather than graph size. RAM is used for caching vertices, edges and properties that have been recently accessed. see this SO thread - the top answer is from Taylor Riggan who is a Principal Graph Architect for AWS Neptune
- For a large number of predicates (>1000), an additional index OSGP can be enabled.
Read Replication
- A single Neptune instance can support up to 15 read replicas. It is best to use read replicas of the same size
Query Languages
AWS Neptune supports Gremlin, OpenCypher and SPARQL.
Neptune supports switching between Gremlin and OpenCypher as required but SPARQL datasets are not interoperable with the other languages.
Indexing Strategy
this page provides a good overview of the differenr indexes used by Neptune and the kinds of queries that each of them can accelerate.