Introducing Data Modeling in the Business Builder

SAP Datasphere has two modeling layers tailored for different user groups. The Data Layer is where data engineers create their models with a technical approach. The Business Layer is for business users who create their models using a more semantic approach. This enables business users to work independently from data engineers, while still being able to collaborate and share data with them. The collaboration between business users and data engineers fundamentally changes: Data engineers can focus on the data consolidation and provisioning, while business users can optimize the business models.

Usually, the IT infrastructure on the data layer is more volatile than business user needs. There are multiple reasons for change, such as system updates, software changes, and acquisitions. On the one hand, this leads to the data layer changing permanently and needing adjustments. On the other hand, business definitions remain relatively stable, for example, they do not often change the KPI margin calculation.

To keep the business layer more robust, while continuously modifying the data layer, the business layer is loosely coupled to the data layer. The layers can be mapped and remapped at any time. Without needing to modify the business models themselves, the data models can be upgraded and remapped to the business models.

Here are some key characteristics of separating IT-driven modeling from business-oriented modeling:

• The separation enables business users to model their business scenario without knowing the underlying data models.
• It provides service modeling for lines of business.
• It hides data complexity through business language and terminology.
• There is one business catalog as the central entry point for business definitions independent of the modeling layer.
• It builds a stable semantic layer for all your consumers.
• When needed, it flexibly changes the underlying data layer (that is, changes in storage technology) without disrupting your consumers.

SAP Datasphere Business Builder is the designated tool for modeling the objects of the business layer. Business users can define business models in the Business Builder, models that are separate from the physical data layer. They create their models top down and map them to the data layer. You can use the business layer to expose business users to the data fields they need while hiding any data fields that might be irrelevant.

Users with the DW Modeler role can use the Business Builder editors to combine, refine, and enrich Data Builder objects and expose lightweight, tightly-focused perspectives for consumption by SAP Analytics Cloud and other BI clients.

In the next picture you can see the different data model types available in the Business Builder.

Each business entity created in the Business Builder consumes data from a Data Builder entity. As you can, at any time, switch the data source of a business entity to a different Data Builder entity, this loose coupling allows you to maintain stable business entities for reporting, even as your physical data sources change.

Business entities can define measures or attributes. Measures are quantifiable values that refer to an aggregable field of the underlying model. An attribute is a descriptive element of a business entity and provides meaningful business insights into measures. The underlying model is a view or a table, which has been created in the Data Builder.

In order to model a meaningful consumption model, business entities define associations between each other. All potential association targets can be pre-defined on the data layer in order to provide business users a variety of modeling options to choose from when preparing their use case-specific consumption model.

Business entities can be modeled as a Dimension or as a Fact. The definition doesn't really differ but rather the intended usage within the Fact Model. Dimensions are generally used to contain master data and must have a key defined. Facts are generally used to contain transactional data and must have at least one measure defined.