Performance and WebSphere Application

Performance is highly dependent on application design and coding, regardless of the power of the runtime platform–a defectively written application will perform just as poorly on z/OS® as it would on another platform. That said, WebSphere® Application Server for z/OS uses the mainframe qualities in hardware, and software characteristics incorporating Workload Management schemes, dynamic LPAR configuration, and Parallel Sysplex® functionality.

Specifically, WebSphere Application Server for z/OS uses the three distinct functions of z/OS workload management (WLM):

Routing
WLM routing services are used to direct clients to servers on a specific system based on measuring current system utilization, known as the Performance Index (PI).
Queuing
The WLM queuing service is used to dispatch work requests from a Controller Region to one or more Server Regions. It is possible for a Work Manager to register with WLM as a Queuing Manager. This tells WLM that this server would like to use WLM-managed queues to direct work to other servers, which allows WLM to manage server spaces to achieve the specified performance goals established for the work.
Prioritize
The application server provides for starting and stopping Server Regions to set work priority. This allows WLM to manage application server instances in order to achieve goals specified by the business.

WLM maintains a performance index (PI) for each service class period to measure how actual performance varies from the goal. Because there are several types of goals, WLM needs some way of comparing how well or poorly work in one service class is doing compared to other work. A service class (SC) is used to describe a group of work within a workload with equivalent performance characteristics.