Use workload management (WLM) classification and define unique service classes for different-priority work running in the same server.
Whenever you are running WebSphere® Application Server on z/OS, you must allow at least one servant for processing each service class. For more information about workload classification, see Classifying z/OS® workload.
Each listener port has one throttle that controls the rate at which differently-prioritized messages are queued as work records on the WLM queue.
When there is a backlog of messages on the WebSphere MQ queue for the message-driven bean, you want certain messages to be processed before others based on transaction class. However the RA listener selects messages from the WebSphere MQ queue and puts them on the WLM queues without considering transaction class. WLM prioritization does not occur while messages are browsed by the queue agent thread – prioritization occurs when work records are queued up.
To ensure that the WLM queue is loaded sufficiently to allow WLM prioritization, set the high threshold (that is, the value of the maximum sessions property of the listener port) higher than the baseline recommendation of "twice the combined number of worker threads in all the servants for the server ."