By using WLM within a sysplex, a TCP/IP stack on z/OS can be configured to direct traffic to the LPAR with the lightest workload.

Now let's take a look at z/OS Workload Manager (WLM). WLM might be described as a performance expectation facility. It can be used to define performance goals (for example, response time) for different applications and different types of work.

In addition, WLM can be used to provide information about the overall workload levels on each image within a sysplex. For more information on sysplex and WLM, see Introduction to the New Mainframe: z/OS Basics.

By using WLM within a sysplex, a TCP/IP stack on z/OS can be configured to direct traffic to the LPAR with the lightest workload. The mechanism is relatively simple from the perspective of the TCP/IP application: TCP/IP asks WLM to provide information on workload levels within each LPAR in the sysplex. WLM responds with a list containing the LPAR name (system name) and a number for reach LPAR. Each number indicates the number of connections that should be sent to each LPAR for this time interval.

In the event that a sysplex becomes fully utilized, WLM can be configured to prioritize workload. In turn, TCP/IP distributes connections to the host with the lower priority workload. Or, if so configured, WLM can also distribute work based upon specific service class goals of the target application's address space. This is referred to as server-specific WLM recommendation.