Cloud applications ingest and process huge amounts of data. Usually by leveraging relational database features or through a message broker.

Streaming architectures give your application tools for at-least-once processing, exactly-once processing, backpressure management, partitioning, joining data, aggregating, and performance metrics generation.

I'll show some basic toy examples that implement the actual features you want to be using. Then compare that to the APIs you can leverage using RabbitMQ or Kafka through Kafka Streams and Spring. In the end you should be able to use streaming instead of reimplementing a worker pool running tasks and adding the same feature by feature each time.