How to Enable Message Partitioning in Spring Cloud Stream for Consistent Consumer Routing

After learning from the previous article about message‑driven microservices, we can ensure that a message is consumed by only one instance in a multi‑instance environment. For scenarios where messages sharing certain attributes should always be processed by the same consumer instance, message partitioning is required.

Configuring Message Partitioning in Spring Cloud Stream

Consumer side

Modify the SinkReceiver application’s configuration to add the following settings:

spring.cloud.stream.bindings.input.group=Service-A
spring.cloud.stream.bindings.input.destination=greetings
spring.cloud.stream.bindings.input.consumer.partitioned=true
spring.cloud.stream.instanceCount=2
spring.cloud.stream.instanceIndex=0

Explanation of the parameters: spring.cloud.stream.bindings.input.consumer.partitioned=true: enables partitioned consumption. spring.cloud.stream.instanceCount=2: total number of consumer instances. spring.cloud.stream.instanceIndex=0: index of the current instance (starts at 0, must be less than instanceCount).

Producer side

In the SinkSender application, add the following bindings:

spring.cloud.stream.bindings.output.destination=greetings
spring.cloud.stream.bindings.output.producer.partitionKeyExpression=payload
spring.cloud.stream.bindings.output.producer.partitionCount=2

Parameter details:

spring.cloud.stream.bindings.output.producer.partitionKeyExpression=payload

: defines the SpEL expression used to compute the partition key from the message payload.

spring.cloud.stream.bindings.output.producer.partitionCount=2

: sets the number of partitions for the destination.

Running the services

Start both the producer and the consumer applications. When multiple consumer instances are launched, assign each a different instanceIndex. With the same message sent to the consumer group, only the instance whose partition matches the computed key will receive and process the message, ensuring consistent handling of related messages.