For my neo4j use case I'm periodically rebuilding the entire DB from CSVs using the import tool. After a rebuild, I need to create two B-tree indexes before the DB can be queried.
This import is running quite fast. With 500M edges it's taking just over an hour. However the index building doesn't seem to be scaling as well.
The neo4j DB is read-only, meaning the only time I insert data is with the import tool. Is there a way to exploit this fact to speed up indexing? Consistency probably doesn't need to be checked in this case
I'm using neo4j:3.5-enterprise