The distribution of changes to material master data is based on ALE change pointers that are generated when master data is created.
ALE standard triggers the distribution of master data using transaction BD21. Subsequent processing is handled in an individual process for all master data that is to be distributed. If a particularly large amount of data is to be distributed, main memory consumption may increae and your system may experience particularly long runtimes.
Report RBDMIDOC_RETAIL can be used to avoid these problems. After the change pointers have been read by the system, the master data changes are sent in data packages. It is also possible to use parallel processing internally to handle the data. In doing so, main memory consumption is reduced. The runtime for the distribution of data is also considerably reduced.
Prerequisites
If you want to use data packages and parallel processing in your system, you are required to create a suitable aRFC Server Group in your system. Use transaction RZ12 to do so. When doing so, you must ensure that all the servers to which data is distributed have sufficient dialog work processes available.
Features
The algorithm for distributing master data functions as follows:
- System reads all unprocessed change pointers for the relevant message type.
- System analyzes data to ascertain the materials for which changes exist. A work list for sending master data is generated.
- System sends generic material including the assigned variants (if any exist). The generic material with the greatest number of change pointers is sent first.
- System sends single materials (if any exist). The single material with the greatest number of change pointers is sent first.
Data packages can contain both generic materials and single materials if data package creation could not be completed for the generic material that was sent last.
The distribution logic ensures that every data package with a genric material also contains all the changes to the variants of the generic material. This, therefore, ensures that the changes to the generic material and its variants can be sent in an IDoc at a later time.
In addition to the distribution of master data changes, the simulation mode enables the user to check the paramaters that have been selected for data package creation, without actually having to send any data. In addition, the user can also select to have a detailed distribution log generated by the system.
To ensure that data packages and parallel processing are used correctly, the user must ensure that the selection parameters for data package creation are correctly configured. The most important selection parameters are detailed in the next section of this document.
Selection
- Frame “Selection of Changes”:
- User selects the message type that is to be used.
- Message Type:
- Only message type ARTMAS or associated message types can be used with this report to distribute master data changes.
- Frame “Configuration” : “Parallel Processing”:
- User can choose to process data in data packages or using parallel processing.
- aRFC Server Group:
- User enters the name of a valid aRFC Server Group. If data package processing without internal parallel processing is to be used for master data distribution, this field can be left empty.
- Number of Parallel Processes:
- If data is to be distributed internally using parallel processing, use this selection parameter to define a maximum number of parallel processes. This limit will not be exceeded when data is distributed even if the server group that is being used does not allow more than the defined number of processes to be used. This means that other processes can be run on the same server group at the same time, without all available resources being used to distribute changes to master data.
- Frames “Configuration” : “Package Processing”:
- Targets (limitation of main memory consumption/parallel processing) can only be achieved most efficiently if the selection parameters for data package creation have been configured appropriately.
- For each logistical material unit (single material or generic material with all relevant variants), the system checks if one of the parameters for data package creation has been exceeded. If this is the case, data package creation is completed and the package is sent. Note that the data package creation process groups different logistical material units into a single package when one of the parameters that is used is actually undercut. The dimensions of these parameters should not, therefore, be too limited. Depending on the parameters that are selected for data package creation, a package can contain both generic materials and single materials.
- Data package creation aims to create exactly the right amount of data packages – not too many but not too few. If too many data packages are created (for example, one data package for each individual material), too many individual RFC (parallel processes) have to be managed and communication requirements become excessive. It could also be the case that the distributing processes cannot distribute the new packages quickly enough into processing. If too few data packages are created, main memory consumption cannot be restricted enough. It may also be the case that larger, individual data packages may dominate the overall distribution runtime despite the majority of the other packages already having been processing in a work list.
- Using data package creation correctly is reliant on the fixed data structure on which master data distribution is based, as well as on the available system resources. It is, therefore, very difficult to make general recommendations. Default values are displayed on the selection screen. The simulation mode also empowers the user to monitor the selected parameters without first having to trigger distribution. This means that no parallel processes have to be triggered, no IDocs are generated and no change pointers are flagged as having been previously processed.
- Number of Materials Per Process:
- Defines the maximum number of materials per process. The variants of a generic material are counted as a single material. As the variants have to be processed in the same data package as the base generic article, the number of materials in the package may actually exceed the value that has been entered by the user.
- Number of Change Pointers Per Process:
- Defines the maximum number of change pointers per process. If this value is exceeded, the data package will contain few materials than the ‘Number of Materials Per Process’. As the variants always have to be processed in the same data package as the base generic material, the data package may contain more process change pointers than the maximum number entered by the user.
- Frames “Configuration” : “Log / Simulation”:
- The system can generate a detailed log for master data distribution. Alternatively, the user can choose to simulate distribution.
- Display Log:
- Log can be output on the screen or in the spool list for the job.
- Simulate Distribution:
- When distribution is simulated, the change pointers are read and data packages are created in accordance with the parameters described above. When distribution is simulated, a log detailing the process is also created. If medium-sized packages are used in distribution simulation, the amount of main memory consumed represents an approximation of the amount of main memory that would be used, were distribution actually to be run.
Output
Report RBDMIDOC_RETAIL provides a detailed log that documents the use of system resource during the distribution runtime. The log is broken down into five sections:
- Distribution Log 1:
- Displays the task number (sequence number), the number of materials and change pointers per process, thus indicating the quality of data package creation.
- Distribution Log 2:
- Displays the task number (sequence number), the start, end and runtime per task, the server and the number of parallel processes during runtime, thus indicating the quality of data load distribution.
- Generic Material Statistics:
- Displays the generic materials sent, the number of variants and the total number of change pointers per generic material and variant.
- Runtime Overview:
- Logs start time, end time and the runtime for the individual steps during processing (for example, read change pointers, structure key tables).
- Data Volume / Parameter Selection:
- Displays the number of materials sent subdivided into generic materials, variants and other materials (for example, single materials) as well as selection parameters (number of materials per process, number of change pointers per process, number of parallel processes).
