Simpler, more reliable and secure.
Neo4j Connector for Apache Kafka and Confluent v6 is out. It’s simpler to configure, more reliable and more secure for event-driven knowledge graph systems built on Neo4j. The blog covers what has changed and what to check before you upgrade; you can also join us at NODES 2026 for a demo.
Why Kafka and Neo4j
Kafka and Confluent are a first-class integration path for Neo4j, and v6 improves the three things most teams use it for.
Real-time change streaming. Change Data Capture (CDC) is available in Aura Virtual Dedicated Cloud, Business Critical, and Neo4j Enterprise. The connector’s CDC strategy publishes change events to Kafka topics with no changes to your data model. That makes Neo4j a solid source of truth for event-driven setups like entitlement sync, governance alerts, and entity resolution. And with a single unified schema in v6, the consumers downstream get simpler too.
Change data capture and Neo4j Connector for Confluent and Apache Kafka Go GA – Neo4j Graph Intelligence Platform
Confluent Cloud. The managed Neo4j Sink Connector supports private endpoints on AWS, Azure, and Google Cloud. You get a fully managed, secure pipeline (with upgrades included) between Confluent Cloud and Aura. Need a source connector, or running on the Confluent Platform? The connector is available in the Confluent Marketplace.
Neo4j sink connector for Confluent available as a managed connector on Confluent Cloud – Neo4j Graph Intelligence Platform
Zero-downtime migration to Aura. The migration guide uses CDC and the connector to keep Aura in sync with a live self-managed database. Restore a backup, replay everything that changed in order with exactly-once semantics, and then temporarily pause writes to cut over. Reads keep working the whole time. v6’s stricter exactly-once handling makes this a safe and robust way to move production databases.
Zero downtime migration to Aura – Neo4j Connector for Kafka
What’s new in v6
The full list is in the 6.0.0 release notes. The highlights:
One schema for CDC events. Node and relationship change events now share a single schema. Whether you sink events into another Neo4j database, feed a stream processor, or land them in a warehouse, you now handle one message shape instead of two. Most CDC pipelines will benefit from this one.
OIDC authentication. The connector can now authenticate with your identity provider rather than relying solely on static credentials. If you standardise on SSO and short-lived tokens, this is for you.
Exactly-once, enforced and automatic. v6 enforces and automatically creates the constraint that exactly-once delivery depends on, so you don’t have to manage. The Aura zero-downtime migration guide relies on the same principle.
Better maps and UUIDs. The new neo4j.query.map-encoding property replaces neo4j.query.force-maps-as-struct and gives you control over how map values are encoded. UUID values are now supported too, so you can stop converting them to strings.
A modern foundation. The connector is built on the latest Neo4j Java driver (6.x), Kafka Connect 4.3.1 and Confluent Platform 8.3 or later. Day-to-day operations benefit from simple improvements: updates now accept null property values, collections with different keys now merge, and only bad record data goes to the dead letter queue.
Breaking changes
Three things to check before you upgrade from 5.x:
- New minimum platform versions. v6 requires Java 17, Apache Kafka Connect 4.3, and Confluent Platform 8.3 or later.
- Sink connectors check the exactly-once offset constraint at startup. Make sure the constraint is in place, or use the new auto constraint mode so the connector handles it.
- Node and relationship CDC events use one schema. If a consumer parses the old separate shapes, update it to read the unified one.
Please note: neo4j.query.force-maps-as-struct is replaced by neo4j.query.map-encoding, so update your config if you use it.
Upgrading and getting started
Check the breaking changes, and review the 6.0.0 release notes for details on all features, fixes, and behavioural changes. The connector documentation covers installation, configuration, and the Confluent Cloud quickstart. There have been substantial changes to the documentation as well, which include the introduction of quick starts that help you select the right sink and source strategies. Back in v5.3, we introduced some monitoring improvements; make sure you are taking advantage of those too.
Join us at NODES 2026
Want to see v6 in action? Join Florent Biville and Vasif Abdulhamidov, Developers at Neo4j, for Neo4j Kafka Connector 6: A Major Step Forward at NODES 2026.
Neo4j Kafka Connector 6: A Major Step Forward
When: 8:30 am to 8:45 am GMT, November 12
Florent and Vasif will cover what changed under the hood and why it matters for building, operating, and observing Kafka pipelines with Neo4j. Register for NODES 2026 and add the session to your agenda.
Neo4j Connector for Apache Kafka and Confluent v6 was originally published in Neo4j Developer Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.
