Real-Time
InfoCubes
Real-time InfoCubes differ from standard InfoCubes in their ability to support parallel write accesses. Standard InfoCubes are technically optimized for read accesses to the detriment of write accesses.
Real-time InfoCubes are used in connection with the entry of planning data. More information:
● BI Integrated Planning: InfoProvider
● Overview of Planning with BW-BPS
The data is written to the InfoCube simultaneously by multiple users. Standard InfoCubes are not suitable for this. You should use standard InfoCubes for read-only access (for example when reading reference data).
Real-time InfoCubes can be filled with data using two different methods: using the transaction for entering planning data, and using BI staging, whereby planning data cannot be loaded simultaneously. You have the option of converting a real-time InfoCube. To do this, choose Convert Real-Time InfoCube from the context menu in your real-time InfoCube in the InfoProvider tree, . In the default setting, Real-Time Cube Can Be Planned, Data Loading Not Permitted is selected. Switch this setting to Real-Time Cube Can Be Loaded With Data; Planning Not Permitted if you want to fill the cube with data using BI staging.
When you enter planning data, the data is written to a data request of the real-time InfoCube. As soon as the number of records in a data request exceeds a threshold value, the request is closed, and a rollup is carried out for this request in defined aggregates (asynchronously). You can still roll up, define and collapse aggregates as before.
Depending on the database they are based on, real-time InfoCubes differ from standard InfoCubes in the way they are indexed. For an Oracle DBMS, this means that no bitmap indexes can be used on the fact table.
Reduced read-only performance is a drawback of real-time InfoCubes, but is outweighed by parallel (transactional) writing option and improved write performance.
When creating a new InfoCube in the Data Warehousing Workbench, select the Real-Time indicator.
Conversion with Loss of Transaction Data
If the standard InfoCube already contains transaction data that you no longer need (for example, test data from the implementation phase of the system), proceed as follows:
...
1. In the InfoCube maintenance transaction in the Data Warehousing Workbench, choose InfoCube → Delete Data Content from the main menu. The transaction data is deleted, and the InfoCube is set to inactive.
2. Continue with the same procedure as when creating a real-time InfoCube.
Conversion with Retention of Transaction Data
If the standard InfoCube already contains transaction data from the production operation that you still need, proceed as follows:
Execute ABAP report SAP_CONVERT_NORMAL_TRANS under the name of the corresponding InfoCube. Due to the increased runtime, schedule this report as a background job for any InfoCubes with over 10,000 data records.
The following typical scenarios arise for the use of real-time InfoCubes in planning:
Actual data (read-only access) and planning data (read-only and write access) have to be held in different InfoCubes. A standard InfoCube should therefore be used for actual data and a real-time InfoCube used for planning data. Data integration is achieved using a multi-planning area that contains the areas assigned to the InfoCubes. Access to the two different InfoCubes is controlled by the automatically added Planning area characteristic.
In this scenario, the plan and actual data have to be together in one InfoCube. This is the case with special rolling forecast variants for example. You have to use a real-time InfoCube, since both read-only and write accesses take place. You can no longer directly load data that has already been added to the InfoCube by means of an upload or import source. To load data despite this, you have to make a copy of the real-time InfoCube and flag it as a standard InfoCube and not as real-time. Data is loaded as usual and is updated to the real-time InfoCube subsequently.