(SP26) Loading data from OutBoard for Analytics

Loading data from one InfoCube (or DSO) into another InfoProvider is a standard procedure in any SAP BW system; this is done by the use of DTPs (Data Transfer Processes) or 3.x Loads. When the source InfoProvider has OutBoarded data, the DTP can read the online as well as from Nearline data.The use of Semantic Groups has beneficial effects on reporting and loading.

To set the extraction form, go to transaction RSA1 and find the DTP you want to edit. 

_{Data Transfer Process in transaction RSA1}

Double-click on a Data Transfer Process. In the "Extraction" tab, you will see an option "Extraction From…" at the bottom. 

_{Displaying a DAP selecting from Active Table (with archive)}

Press  to edit, switch to change mode,then select where you wish to extract data from. "Archive (Full Extraction Only)" is available only if "Extraction Mode" is "Full".

_{Extraction from}

To improve the performance of loading process, reading can be parallelized. This behavior is set by OutBoard parameter LOAD_PAR_WP. (This can be changed in the settings of the Outboard Cockpit) If set to > 0, when data for loading is requested, then the request is split into groups consisting of multiple archiving data packets and read by reading processes. The maximum number of data packets in a group is defined by the parameter LOAD_MAX_GRAN (Settings of the Outboard Cockpit).
The main process starts reading processes as background work processes. When the reading process is finished for a group, it saves the data in the storage area. Subsequently the main process gathers the results, sends them to DTP process and starts a new reading process. To prevent the piling up of data in the storage area, LOAD_PAR_RET (Settings of the Outboard Cockpit)  parameter is used.
This parameter determines maximum number of finished processes. If the number is higher, no new reading processes are scheduled. If parameter LOAD_PAR_WP is set to zero, no parallelization is used. In this case, value of LOAD_PAR_RET is irrelevant. The parameter can be set on InfoProvider level.

Bear in mind that also writing of data can be (and usually is) parallelized.By calling the report /DVD/NLS_DTP_FLAG_SETUP a user can specify the list of DTPs for which the data from NLS should be included in the loading. Only DTPs with ODS sources can be modified by this report. 

_{Include NLS Data in Loading automatically}

The following parameters can be specified:
 Choose ODS's DTP:
          All DTP's – the Archive flag will be updated for all DTPs on the system with ODS Source.
          All DTP's for Source with DAP – the flag will be updated only for DTPs with ODS sources for which the DAP was already created.
          Manual Selection of DTPs – the user selects the DTPs manually.
Request/task for DTP: In case a DTP is not assigned to a request, it will be assigned to the request selected here. The selected request should obtain a task for the user running the report.
Delta Init. Extraction Form: determines, if data from archive is included in load or not.
Even if the DSO can be archived, there can be some limitation for DTP extraction from the archived DSO. For example:

• If using navigational attributes in the DTP extraction (not possible or restricted upon archiving) or
• If using semantic grouping in the DTP extraction (data having the same semantic key can be extracted multiple times – first from active DSO table, second time from DSO NLS area).

 These cases, the DTP logic should be adjusted or the usage of these specific extraction scenarios should be avoided. The workarounds for these scenarios are:

• If using navigational attributes in the DSO extraction: The extraction from 7.x data source can be used instead.
• If using semantic grouping in the DTP extraction: The logic of semantic grouping should be moved to the start routine.