Imagine you have a transaction that goes through several services. Each of them performs some logic and results in a decision about the transaction. In certain cases, you need to aggregate several decisions in order to come up with a more detailed one, and eventually, you need to resolve all the different decisions and choose the right one.
When trying to figure out the best way to solve this problem, we looked into Kafka streams. An application that allows you to easily aggregate Kafka events from one or multiple topics.
In this talk, I will present how we used Kafka streams in production to combine events from different services and come up with an aggregated decision for a specific transaction, and how we handled windowing on our events to create the most accurate decision.