Why SAC and Datasphere Are Integrated: Purpose and Architecture Overview
Understand the business and technical reasons SAP pairs SAC with Datasphere, and how the two products divide responsibility for data modeling versus visualization and planning.
Explanation
SAP Analytics Cloud (SAC) and SAP Datasphere are two distinct but deliberately complementary cloud products, and understanding why they exist together is the foundation for every later configuration decision. SAC is a business intelligence, planning, and predictive analytics tool: it renders stories, dashboards, and planning models, and it runs calculations against data at query time. Datasphere (formerly SAP Data Warehouse Cloud) is a data management and modeling layer: it connects to source systems, replicates or federates data, harmonizes it into semantic layers, and exposes clean, governed data models for consumption. The integration exists because most enterprises need both capabilities but do not want to duplicate modeling logic in two places. Before Datasphere existed as a separate product, SAC had its own live/import connections to various source systems (S/4HANA, BW, HANA, generic ODBC/JDBC sources). Datasphere formalizes and centralizes that modeling work: instead of every SAC story author building ad hoc calculations against raw tables, a data engineer or architect builds a certified, reusable semantic model in Datasphere once, and many SAC stories, planning models, and even other consuming tools can reuse it. This separation of concerns mirrors a classic warehouse-versus-presentation-layer pattern that experienced BW or BI consultants will recognize, but implemented entirely in the cloud. There are two fundamentally different ways SAC consumes Datasphere content, and this is the first architectural decision every project must make. The first is a live connection, where SAC does not store data locally; every query, filter, and drill-down in a story is pushed down as a live query to Datasphere, which in turn may federate to underlying source systems or query its own in-memory tables. The second is an import connection (or acquired/import model), where SAC actually copies a snapshot of data out of Datasphere into SAC's own in-memory model. Each approach has very different implications for data freshness, performance, security model, and license/consumption cost, which later lessons cover in depth. Organizationally, this integration also reflects a governance shift. Instead of allowing every business analyst to import raw extracts into SAC directly from ECC, S/4HANA, or flat files, many enterprises now mandate that all analytical data must first pass through Datasphere, where naming conventions, data quality rules, currency conversion, and row-level security are enforced centrally. SAC becomes purely a consumption and storytelling layer. This is not universally enforced, and some organizations still allow direct SAC import connections to systems like S/4HANA, but the trend in reference architectures is toward Datasphere as the mandatory semantic gateway. A critical distinction to internalize early: Datasphere spaces are the unit of workspace isolation and security, and SAC connects to specific spaces, not to the whole Datasphere tenant. Within a space, Datasphere exposes 'shared' objects for outbound consumption; only artifacts explicitly marked or exposed for consumption (depending on the object type and space configuration) are visible to SAC when building a connection. This means a common early confusion is a consultant building a perfect view in Datasphere and then not finding it in SAC's connection browser, because it was never exposed for consumption from that space. Finally, it's important to understand what this integration is not. Datasphere is not a replacement for SAC's planning engine, and SAC is not a data warehouse. Attempting to do heavy ETL-style transformation logic inside SAC calculated dimensions, or attempting to build interactive story-like dashboards inside Datasphere, both go against the intended architecture and typically create maintenance and performance problems later.
Real project scenario
A retail company's finance team previously had five different SAC models all recalculating the same currency conversion and cost-center hierarchy logic independently, leading to reports that disagreed with each other in monthly reviews. During a platform modernization project, the BI architect introduced Datasphere as the single semantic layer: one Datasphere space was built for finance, containing a certified analytic model with currency conversion, hierarchies, and row-level security applied once. All five SAC stories were migrated to consume that one model via live connection, and the recurring 'why do these two reports not match' escalations stopped within the first reporting cycle.
Common mistakes
• Assuming SAC can browse any table or view in Datasphere automatically, without realizing objects must be explicitly exposed for consumption from the space • Treating Datasphere as just another data source identical to a plain HANA database connection, ignoring its space-based security model • Building transformation logic redundantly in both SAC calculated dimensions and Datasphere views, causing inconsistent results • Not deciding early between live and import connection strategy, leading to a costly re-architecture mid-project • Assuming all Datasphere capabilities are identical across all customer tenants, when feature availability can vary by release and licensing tier
Best practices
• Establish Datasphere as the single semantic layer for shared metrics before building multiple SAC stories against raw sources • Always confirm which objects are explicitly exposed for consumption before assuming SAC will see them • Document the space and connection type (live vs import) used by each SAC model as part of standard project handover artifacts • Avoid duplicating business logic across both tools; decide ownership of each calculation explicitly • Review Datasphere space design with the SAC modeling team early, not after story development has started
Interview angle
Interviewers commonly ask candidates to explain the conceptual difference between SAC and Datasphere and to justify when to use live versus import connections; a strong answer references the separation between the modeling/warehouse layer and the presentation/planning layer, and explicitly mentions space-based exposure for consumption as a gating factor, rather than giving a generic 'one is BI and one is a warehouse' answer.