Message Flow in a Composite Cluster Connection
According to the governing policy, a composite cluster connection distributes the messages to the child cluster connections. The child cluster connections then distribute these messages using its load balancing policy to the respective Brokers.
For example, consider the following composite cluster connection factory configuration:
A composite cluster connection factory contains two child cluster connection factories: Child cluster connection factory 1 and Child cluster connection factory 2.
The composite cluster connection factory is configured for failover using the round robin load balancing policy.
Child cluster connection factory 1 is configured to use random load balancing policy.
Child cluster connection factory 2 is configured to use sticky load balancing policy
The following figure illustrates this process.
1. The first message is published to child cluster connection factory 1 based on the round robin load balancing policy.
2. The child cluster connection factory 1 routes the message to Broker A, or Broker B, or Broker C based on the random load balancing policy.
3. The second message from the publisher will then be published to child cluster connection factory 2.
4. The child cluster connection factory 2 sends the message to Broker X based on the sticky load balancing policy.
5. Third message is published to child cluster connection factory 1 and so on.
6. If all the Brokers A, B, and C of child cluster connection factory 1 goes offline, all subsequent messages are routed to the child cluster connection factory 2.
