Standalone vs Distributed Connect

A developer needs to test a connector against a sample inventory file on a laptop. A production team needs a database integration that can resume on another machine when its current host fails. Both jobs can use Kafka Connect, but they need different ways to manage execution and recovery.

Kafka Connect provides two execution modes: standalone and distributed. The choice determines where Connect keeps its management state, how it places work, and what happens when a worker disappears.

In this chapter, we'll compare those modes using Kafka Connect 4.3 and a KRaft-based Kafka cluster. We'll separate the behavior Connect supplies from the storage, deployment, and connector capabilities needed to make recovery work.