Adopting Spring Cloud Stream

KakaoPay's Information Platform Team is utilizing Spring Cloud Stream and Kotlin for the task of combining various service data and submitting it to external institutions.

The previously used spring-kafka had drawbacks such as being a low-level development approach that made it difficult to enforce coding style, made data flow hard to grasp, and increased coupling between code. In contrast, Spring Cloud Stream is advantageous for reducing coupling since it allows standardization of data flow control and coding style.

When implementing data extraction logic, each source—such as the payment API, points API, and Hadoop—is defined as an individual Function bean to compose the pipeline. This allows flexible response when new data sources are added or existing logic changes, and makes it easier to separate resources and scale up each stage.

The Binder connects the application to message brokers like Kafka or RabbitMQ, and thanks to the abstracted structure, switching brokers is also simple.