neo4j-kafka-skill

To stream Neo4j changes to Kafka reliably, configure the Kafka source connector using the Neo4j native CDC query API, which captures graph modifications and publishes them to Kafka topics with exactly-once semantics and dead-letter queue error handling.

You can write streaming data into Neo4j using Cypher, Pattern, CDC (schema/source-id), and CUD message formats. These sink strategies map Kafka messages into graph writes by transforming or applying streaming events directly to the graph database.

Yes, you can consume Neo4j change data capture events without Kafka by using the native db.cdc.query API. This in-process CDC consumption approach allows you to query change events directly within Neo4j without requiring a Kafka Connect pipeline.

Exactly-once semantics with the Neo4j Kafka Connector work by using EOS offset tracking constraints. This ensures that streaming events between Kafka and Neo4j are processed exactly once, preventing duplicate graph writes during connector failures or rebalances.

Yes, the Neo4j Kafka Connector supports Kafka Connect configuration including schema registry converters. This allows you to integrate schema validation and serialization formats when mapping Kafka messages into graph writes or publishing graph changes.

You handle failed messages when streaming data between Kafka and Neo4j by configuring a dead-letter queue (DLQ). This production-grade error handling mechanism captures unprocessable events, allowing the pipeline to continue streaming without blocking valid graph writes.