z/OS MVS Programming: Workload Management Services
Previous topic | Next topic | Contents | Index | Contact z/OS | Library | PDF


Modelling Your Business Transactions

z/OS MVS Programming: Workload Management Services
SC34-2663-00

Modelling Your Business Transactions

With the introduction of support for EWLM, there are two ways of modelling your business transactions:

  • Using enclaves without EWLM participation
  • Mapping EWLM work requests, either closely or loosely, to enclaves

Using Enclaves without EWLM Participation

With this model, a business unit of work is represented by the lifetime of the enclave under which it will be processed, that is, from the time the enclave is created until it is deleted. At the time the enclave is created, the business unit of work is classified and assigned a WLM service class. The work manager does not further interact with EWLM to identify when such a business unit of work starts and stops, and the work manager is not an EWLM participant.

Mapping EWLM Work Requests to Enclaves

With the EWLM model, a business transaction is measured by the EWLM ARM agent and may be referred to as a work request. A work request lasts from the time it is started until the time that it is stopped. Only EWLM participants can start and stop work requests. Work requests are mapped to an enclave, as the enclave is the MVS™ means to manage (with SRM) transactions directly across address space boundaries.

Work requests may be started and stopped explicitly by the EWLM-participant application. In this case, the work requests can be said to be loosely mapped to enclaves. Alternatively, work requests may be started implicitly when the enclave is created, and stopped implicitly when the enclave is deleted. In this case, the work requests can be said to be closely mapped to enclaves. Note that such work requests do not differ from the non-EWLM model, except that:

  • EWLM is notified about the work request’s start and stop time
  • EWLM may optionally classify the work request based on its end-to-end policy

It is also possible to combine the explicit and implicit starting and stopping of work requests. For example, a work request my be started explicitly but stopped implicitly. The possibilities, along with the associated WLM services, are shown in Table 13.

Table 13. Starting and Stopping Work Requests
StartWLM ServiceStopWLM Service
ImplicitIWM4ECREImplicitWork request is stopped by IWM4EDEL
IWM4ECREExplicitIWMESTOP
ExplicitIWM4ECRE with ESTRT=EXPLICIT and subsequent IWMESTRTImplicitWork request is stopped by IWM4EDEL
IWM4ECRE with ESTRT=EXPLICIT and subsequent IWMESTRTExplicitIWMESTOP
ExplicitIWM4ECRE with ESTRT=EXPLICIT_SINGLE and subsequent IWMESTRTImplicitWork request is stopped by IWM4EDEL
IWM4ECRE with ESTRT=EXPLICIT_SINGLE and subsequent IWMESTRTExplicitIWMESTOP

An EWLM participant can process multiple single work requests under the same enclave. The underlying assumption is that an enclave can only process one work request at a time. Within a work request, however, processing of sub-work requests is allowed. The ARM transaction model requires that nested sub-work requests be completed before the nesting work request completes. WLM enforces this requirement.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014