The solution to this is to make it possible to specify a key for each message that the messaging system can use in order to guarantee that the order is kept between messages with the same key. This can be solved by introducing sub-topics, also known as partitions, in a topic. The messaging system places messages in a specific partition based on its key. Messages with the same key are always placed in the same partition. The messaging system only needs to guarantee the delivery order for messages in one and the same partition. To ensure the order of the messages, we configure one consumer instance per partition within a consumer group. By increasing the number of partitions, we can allow a consumer to increase its number of instances. This increases its processing message performance without losing the delivery order

In Spring Cloud Stream, this needs to be configured on both the publisher and consumer side. On the publisher side, the key and number of partitions must be specified. For example, for the product-composite service, we have the following:

https://medium.com/latentview-data-services/how-to-use-apache-kafka-to-guarantee-message-ordering-ac2d00da6c22