Ansicht
Dokumentation

CRM_MKT_ELM_PARALLEL - Define Settings for Parallel Processing

rdisp/max_wprun_time - Maximum work process run time   Vendor Master (General Section)  
This documentation is copyright by SAP AG.

In this activity you define parallel processing for the process steps in a list.

The Parallel indicator determines whether the data in the individual process steps is processed in parallel or sequentially. Settings from other system parameters are only used if you set this indicator.

In parallel processing the data is put together into packages. An asynchronous RFC (aRFC) is started for each package. The package data is processed in this aRFC. The aRFCs run in dialog work processes.

The server group defines the group of application servers that should be available for the dialog work processes for parallel execution. Using server groups has the advantage that the system uses the current system load as a basis for selecting the instance to be used. When starting an aRFC with a server group entry, the instance is not determined until the point when a dialog work process is used for execution. All instances belonging to the group are then taken into account.

The maximum number of aRFCs defines how many dialog work processes from the server group are used by parallel running aRFCs.

Determining the maximum number of aRFCs

By using a larger number of parallel processes, the throughput can be reduced. This can be because of lock conflicts (ENQUEUEs or database blocks) or overloading of system components (CPU, main memory, ENQUEUE server overload, database IO). Peak overloading of the system when there is intensive parallel processing is therefore higher than in parallel, serial execution. If the capacity limit of one of the system components above is reached, this restricts the throughput of the whole process. When you start further processes, this reduces the throughput.

The exact definition of the process number is project-specific and depends on the available system resources. There is no general rule that applies. It is highly recommended that you test the exact settings by doing a stress test.

At the start of parallel execution the required server group is initialized and the dialog work processes that are currently available are determined. If the number of dialog work processes currently available on the server group are smaller than those given by the user, this is used as a guideline for execution. Otherwise the system works with the number entered by the user.

Entering a server group

The definition of a server group with instance assignment and the temporary definition of value limits for the instance load situation can be carried out using transaction RZ12. The F1 help describes the meaning of the individual parameters.

When you enter the server group, you should pay attention to capitalization. To avoid errors, use the F4 help.

Almost all parameters in RZ12 are given in percentages. The exceptions are: Parameter for generally switching on and off value limits, Parameter, for defining the minimum number of free dialog processes and parameters for the maximum waiting time after the load check in the system. For every instance there is also only one record of valid parameters. You cannot define different values in different server groups for an instance.

Experience has shown that the parameter for the minimum number of free dialog processes should be set to >= 3 and that more dialog work processes should be offered via the server group than are actually used. This enables you to better deal with unexpected load peaks or to minimize the danger of temporary resource bottlenecks.

ROGBILLS - Synchronize billing plans   ROGBILLS - Synchronize billing plans  
This documentation is copyright by SAP AG.

Length: 3894 Date: 20240523 Time: 173115     sap01-206 ( 99 ms )