How best to do this depends on a lot of factors including what other systems you have in place, how much data, how frequently it gets updated, and so on.
A common approach to suggest as a starting point:
- Dump your existing relational DB as CSV and then use neo4j-import or LOAD CSV to do an "initial load" that gets most of it in.
- Set up change data capture (CDC) from your relational database, for example to publish all changes to Kafka, and then use the neo4j-streams plugin to subscribe to those changes going through kafka. This handles the "ongoing" part of the change.
What you end up with is a Neo4j instance that is a follow-along copy of what's happening in the RDBMS.