Show TOC

Real-Time Data AcquisitionLocate this document in the navigation structure

Description

Real-time data acquisition (RDA) supports operative reporting. The data is transferred to the BW at regular intervals and is then updated to the DataStore objects, which are directly available for reporting and analysis. Background processes (daemons) in the BW system initiate the InfoPackages and data transfer processes assigned to them (to update the PSA data to DataStore objects).

To simplify the processing of and reporting on the data, the requests are kept open in the DataStore object and in the PSA. Based on settings in the InfoPackage, the daemons decide when a request is to be closed and when a new request is to be opened.

The data can be transferred from the source to the entry layer in BW (PSA) in two ways:

  • Using a Web Service Push

    A Web service push can write the data directly from the source to the PSA. The data transfer is not controlled by BW. An InfoPackage (for full upload) is required only to specify request-related settings for RDA.

  • Using a Service API

    If the data is transferred via a delta queue of an SAP source system, the data is also written to the PSA. This requires the creation of a special InfoPackage; this InfoPackage ensures data is transferred from the source system and is used for specifying request-related settings. To generate an entry for the DataSource in the delta queue, an InfoPackage with delta initialization must first be created and executed.

    PrerequisiteThe DataSource has to support real-time data acquisition. The option to support real-time data acquisition is a property of the DataSource. From a technical point of view, this property is possible if the BW Service API in the SAP source system has at least the following release: Plug-In-Basis 2005.1, or for 4.6C source systems Plug-In 2004.1 SP10.

    DataSources used for RDA can no longer be used for standard extraction (scheduling using InfoPackages). This is because the DataSource can have one extraction mechanism only (RDA or scheduled data transfer). It is not possible to have multiple extraction mechanisms simultaneously, since the delta queue can only contain one entry for each DataSource and target system at any given time.

For more information, seeReal-Time Data Acquisition.

General Practices with Migration

If you want to integrate the extraction of data with real-time data acquisition into an existing data flow, there are two alternatives:

  • Using two different DataSources

    Up until now, you evaluate the information at item level using one DataSource, for example. To do this, data is made available periodically using a scheduled extraction; InfoPackages specify the selection parameters and control the data request.

    If you want to obtain a more efficient overview of new documents, you can connect a second RDA-enabled DataSource in parallel, to write the header information about the daemon to a separate DataStore object at shorter intervals. The existing data flow does not change. For reporting and analysis, the data can be combined using a MultiProvider (for more information, seeMultiProvider. To avoid redundant data in the RDA DataStore object, regularly delete the data from the DataStore object once the standard DataStore object is loaded.

    The following figure outlines these two options:

    Setting the data flow in this way allows you to continue to use the existing data flow. You can use a separate and possibly smaller DataStore object for the data from the additional DataSource. However, you require a MultiProvider for reporting. Additional administration effort is demanded by the need to regularly delete the data from the RDA DataStore object.

  • Using a single DataSource

    If just one DataSource is to be used, the existing data flow must be fully switched over to real-time data acquisition. This is a result of the fact mentioned above that a DataSource can have one extraction mechanism only. In our example, the information about new documents would not be transferred using a separate DataSource for header information; instead, the DataSource that returns the data at item level would be switched over to real-time data acquisition.

    The following figure outlines these two options:

    If the existing data flow is fully replaced by a new RDA data flow, the data reconciliation for different DataSources does not require any additional administration effort. Switching to this data flow allows you to load larger volumes of data by dividing the load process into several smaller load processes.