Not sure if you're being factitious, but it sounded like parent's data was already in Kafka. But you're right, no reason not to start with kinesis in the first

Not sure if you're being factitious, but it sounded like parent's data was already in Kafka. But you're right, no reason not to start with kinesis in the first place, especially with all the AWS options around it including Kinesis Analytics (Streaming "SQL"), Flink (supported by AWS now) or plain old EMR (Spark, Flink, etc...). There is a lot of work that goes into maintaining your own Kafka ecosystem, even if you use Confluent, Lenses, etc...