Flat InfoCube
A flat InfoCube is built in the same way as the SAP HANA-optimized InfoCube. It is available for databases that support this flat format (see Prerequisites).
Flat InfoCubes only consist of one fact table and the table for the package dimension. The SIDs (master data IDs) of the characteristics in the dimensions are saved directly in the fact table. Therefore no dimension tables are needed, apart from the table for the package dimension. This improves system performance when loading data. Since dimensions are omitted, no DIM IDs (dimensions keys) have to be created.
The dimensions are used simply as a sort criterion and provide you with a clearer overview when creating a query.
Flat InfoCubes can be compressed. For the compressed data in the fact table, the request number is set to null.
Flat InfoCubes can also contain non-cumulative key figures. The procedure is the same as for the SAP HANA-optimized InfoCube. More Information: SAP HANA-optimized InfoCube
Prerequisites
Your SAP BW system is running on a database with at least one of the following database versions: IBM Db2 10.5 with Db2 BLU Acceleration, IBM Db2 for i 7.1, Microsoft SQL Server 2014 or Oracle Database 12c In-Memory (Patch Set 1 (12.1.0.2)).
Converting InfoCubes
You can convert standard InfoCubes into flat InfoCubes using the program RSDU_REPART_UI. InfoCubes containing data can also be converted. Flat InfoCubes can also be converted back to standard InfoCubes.
In the Data Warehousing Workbench, you can call the conversion from the context menu for your InfoCube.
If you want to convert a large number of InfoCubes, you can also use program RSDU_IC_STARFLAT_MASSCONV for mass conversion of InfoCubes.
Converting Semantically Partitioned Objects
You can also convert semantically partitioned objects into flat objects if they are made up of InfoCubes. For more information, see SAP Note 2711358 
.