SAP Data Warehouse Cloud (DWC) offers a wide range of options for building data models that are subsequently used for reports in SAP Analytics Cloud (SAC). Performance is an essential factor, especially for larger and extensive data models. It plays a role at the latest when a story is built in SAC based on the data model from SAP DWC and the loading times of the various tables and graphics are very long.
To improve the performance of the data models, SAP Data Warehouse Cloud provides various options. Optimized query times can be achieved with different join cardinalities for the views as well as with associations, where entities are intelligently linked.
With the help of SQL or SQL script, users can create views in SAP Data Warehouse Cloud. The views are joined from tables or individual table columns. Thanks to the in-memory technology of the HANA database, the calculations and mappings of the various logics run on the fly, so there is no need to duplicate data sets. This enables reporting with real-time data.
The graphical join is aimed at users who want to build a data model in a visualized form and do not want to use coding. Such a view impresses with its good clarity, which makes it easier to understand which tables and table columns make up the view. The view has the same functions as the SQL join.
Associations are able to intelligently link relationships between different entities in SAP Data Warehouse Cloud. This option is available in both SQL view and graphical view. SAP DWC automatically detects whether a relationship can be established between the different entities and executes this association. The intelligent linking of entity relationships contributes significantly to faster performance of the data model.
In addition to performance, good usability of views is also important for modeling. Against this background, associations are the first choice. With them, users can see at first glance which dimensions originate from another table.
In view of the criteria mentioned, it can be summarized that associations should preferably be used for modeling. This is because they offer advantages in terms of both usability and performance. If an association is not possible, joins with corresponding cardinalities should be used. This also achieves excellent results in terms of performance.