Architected Data Mart Layer in the LSA++

Concept

In the LSA++, the data propagation layer consists of SAP HANA-optimized InfoCubes and/or SAP HANA-optimized DataStore objects (in the Business Transformation layer).

SAP HANA-optimized InfoCubes offer the following advantages:

  • Data can be loaded faster because no DIMIDS are required.

  • Data modeling is simpler. Multidimensional modeling is not necessary because the dimensions are only used as structuring criteria and do not have any effect on system performance. Aggregates and DB indexes are not required.

  • The data model can be changed more quickly because it does not contain any data.

Using InfoCubes

Since it is possible with the LSA++ to execute queries directly on the data propagation layer (on the DataStore objects), this raises the question of whether the architected data mart layer and its InfoCubes is actually needed. Note the following criteria:

An InfoCube only accelerates queries if

  • the content of the InfoCubes only contains a subset of the data from the DataStore object (for example, due to a selection in the DTP).

  • the InfoCube only contains information that is available in the DataStore object.

  • no other additional checks are running when data is loaded into InfoCube (for example, referential integrity).

If these criteria have been met, you can delete these InfoCubes. This easiest way of doing this is in a scenario with a MultiProvider on the InfoCube on which the queries are executed. Then you can simply replace the InfoCube with the DataStore object.

Before deleting the InfoCubes, check

  • if the query performance on the MultiProvider on the DataStore object is as good as on the InfoCube.

  • if SID generation is activated on the DataStore object.