We have a uniform distributed queue configured for a JMS port for web services on one domain (Domain1). The Connection factory used for creating the JMS connections has both Server Affinity and Load Balancing enabled.
SAF has been configured on the client Domain(Domain 2). So the web service request is routed through the SAF imported destination in the client domain. No delay has been configured on the SAF module. And we are not doing any batching at this point.
We are observing an un-even load balancing on the UDQ on Domain 1.
We are on WLS 9.2.3.
-managed server 1 on host1
-managed server 2 on host1
-managed server 3 on host2
-managed server 4 on host2
In the above topology, we are observing that managed servers 1 and 2 have about 10K requests and 3 and 4 have 2K requests.
What are the factors influencing load balancing in this case? Note that all the managed servers are started pretty much at the same time.
How does Server Affinity play a role in this case - in the context of SAF, the destination is 'always' remote.