In the SAP BW system on traditional database, a cube has two fact tables – F and E.
When the requests are compressed, the data in the F fact table moves to the E fact table and the request ids are set to 0.
In an SAP BW system on HANA database, a cube does not have two fact tables.
Instead it has partitions to store the compressed and uncompressed requests.
During compression, the system moves the requests from one partition to another and also sets the request ids to 0.
There are two other partitions but they are relevant only for non-cumulative infocubes.
Note: Compression improves loading performance and makes for faster aggregate rebuilding. (Aggregates are not relevant for SAP BW on HANA).
The reduction ratio of the number of records in the F and E fact tables depend on the granularity of data. For example, if the infocube has order number, ocalday and material characteristics and frequency of loading is once a day, there won’t be much reduction in records after compression.