Either would work, the Kafka connector would be my suggestion.
Oracle connector/JDBC Pros:
- Write SQL/cypher directly
Oracle connector/JDBC Cons:
- Need to figure out how to schedule those things to run in Cypher with APOC periodic
- Neo4j server must have extra JARs for Oracle; must do extra work of JDBC integration
- If you need to transform data coming from Oracle, all of this must be done in cypher or SQL
Kafka Connector Pros:
- If you publish oracle data -> Kafka, Neo4j can benefit but other later systems might as well depending on what you need
- Kafka can be used to transform messages on the wire so you don't have to write transforms in Cypher
- If you already use Kafka & are familiar with it, all-around nicer data integration approach
Kafka Connector cons:
- You gotta run Kafka
- The connector itself (kafka connect) is another thing to run, which you might not want to do unless you already have other kafka connectors running
