If I do not want to persist my transactional data in a graph database ( as it is already persisted in a RDB) but I want to use the power of graph databases to get valuable insight from the data. Is there an in-memory version of neo4J that I can use to create nodes and relationships on the fly and query the graph to detect hidden patterns in my data? After the query results are obtained, the graph is no longer necessary.
If there is such a solution available is it production ready, can it scale in a multi-tenant environment?