...

For several years, we have been developing a new way to run Kafka with self-managed metadata. This new mode, called KRaft mode, addresses many urgent scalability and performance issues in Kafka. It also eliminates the need to run an Apache ZooKeeper cluster alongside every Kafka cluster. See KIP-500 for more details.

In Kafka 2.8, KRaft mode was available for "released in early access" for anyone who wanted to test it. By Kafka 3.0, it was in preview. Now, we would like

This KIP proposes to mark KRaft as production-ready for new clusters in the next release, upcoming Kafka 3.4 release. We would also like to deprecate ZooKeeper mode. This will allow us , and plan to remove ZK mode entirely it in the Kafka 4.0 release of Kafka.

Rationale

Over the last year and a half, we have gained a lot of confidence in the KRaft implementation. Our JUnit test coverage has grown as we have converted existing tests and written brand new tests. We have added support for KRaft mode to Kafka's "ducktape" system test framework. We have also run load tests and stress tests on KRaft clusters.

...

The last omission (upgrade from ZK) is the most significant. We expect to release a KIP soon describing in detail how the upgrade from ZK mode will work. This work is currently targetted at the next release (3.4).

Compatibility, Deprecation, and Migration Plan

...