Floodplain is meant to run inside a container, and probably under some kind of container orchestrator like Kubernetes.
Still, running Floodplain is pretty easy. Starting a floodplain instance is just a method call. No frameworks are required, and because of that, it is easy to integrate in other frameworks like Quarkus.
A thing to keep in mind: Floodplain does not run any Kafka Connectors, it just creates configuration objects for a Kafka Connect instance that needs to contain all the required connectors and dependencies.
We’re still unsure how to build that effectively. At the moment we’re still deciding if we would prefer multiple connector instances (So we can use immutable off-the-shelf container instances of our connectors), or try to build a ‘super connector’ that contain all required dependencies.
TBD, I guess.
TODO note: Currently researching Quarkus, possibly combined with GraalVM native.