It sounds like you have a “temporary database” that you are making a lot of requests towards and want to make it available to multiple Flows? I don’t see how you can make it shared in-memory without persisting because each Flow is a separate process. Results are passed from sub-flows to main flows through persistence because they are on separate executors.
Are you transformations OLAP and SQL-like? Cuz I would look into
DuckDB. If it’s more transaction-heavy, this is very hard