使用jdbcIO与Java驱动程序将数据流输出到neo4j

Question

I'm currently working on a setup that uses Google Cloud Dataflow to transform data and save into a Neo4j Database hosted on a Compute Engine VM. The current setup uses a JdbcIO to write to Neo4j by running a prepared statement, but it seems also possible to use a Neo4j driver directly in Java, which allows more flexibility on creating the query dynamically.

I wonder if anyone had compared the 2 approaches and notice any differences? I would guess the Jdbc approach would be more efficient since it only need to run the prepared statement, but that is just my guess.

Update I'm going to post my findings as I experiment more with the two methods

Answer 1

One difference I found was that when using JdbcIO, I run into a deadlock when multiple transaction is trying to write to the same node at the same time. However that issue did not occur when using Neo4j driver directly.