Skip to main content

Module 8: Advanced Data Modeling (BW/4HANA 2.0)

Advanced data modeling in BW/4HANA is about designing for performance, reuse, and flexibility, not just storing data.

This module covers:

  • Star schema vs LSA++ modeling
  • Semantic layering concepts
  • Reusable modeling strategies
  • Performance-oriented design
  • BW/4HANA best practices (real-project driven)

1. Star Schema vs LSA++ (Very Important)​

1.1 Classic Star Schema (Traditional BW Thinking)​

Concept​

  • Central fact table (InfoCube)
  • Surrounding dimension tables
  • Pre-aggregated data
  • Heavy persistence

Characteristics​

  • Strong coupling between fact & dimensions
  • Performance via aggregates
  • Rigid structure
Dimension → Fact (InfoCube) → Query

Limitations in BW/4HANA​

Why Star Schema is No Longer Ideal

Not flexible
High redundancy
Poor adaptability
Over-persistence

1.2 LSA++ Modeling (Modern BW/4HANA)​

Concept​

  • Layered architecture
  • Clear separation of responsibilities
  • Combination of persistence + virtualization
  • Heavy HANA pushdown
aDSO (facts)
   ↓
CompositeProvider (semantic join)
   ↓
BW Query

Advantages​

AreaBenefit
FlexibilityHigh
ReuseStrong
PerformanceHANA-optimized
MaintenanceLower
Interview Line

LSA++ replaces star schema rigidity with layered semantic modeling.

2. Semantic Layering (Core BW/4HANA Principle)​

What is Semantic Layering?​

Semantic layering means:

  • Each layer has one clear responsibility
  • Business meaning is added gradually
  • Logic is not duplicated

Typical Semantic Layers​

LayerResponsibility
AcquisitionTechnical data capture
HarmonizationBusiness logic
ReportingVirtual combination
ConsumptionPresentation
Key Rule

Semantics increase as you move upward.

3. Reusable Modeling Strategies​

3.1 Reuse at InfoObject Level​

  • Global characteristics
  • Shared master data
  • Central attributes & texts
warning

Never create duplicate business dimensions.

3.2 Reuse at aDSO Level​

  • One aDSO = one business meaning
  • Avoid report-specific aDSOs
  • Central harmonized fact layers

3.3 Reuse at CompositeProvider Level​

  • Reusable semantic joins
  • Avoid one CompositeProvider per query
  • Build domain-based providers
Best Practice

CompositeProviders are semantic APIs for reporting.

4. Modeling for Performance (Critical)​

4.1 Pushdown First Strategy​

  • Calculations in HANA
  • AMDP transformations
  • SQL-based joins
warning

If logic runs in ABAP, you already lost performance.

4.2 Persistence Strategy​

Persist When:​

  • History is required
  • Data volume is large
  • Reprocessing is needed

Virtualize When:​

  • Data is reused
  • Joins are simple
  • Real-time access is needed
Golden Rule

Persist facts, virtualize semantics.

5. Key Performance Design Areas​

aDSO Design​

  • Minimal field list
  • Correct key definition
  • Avoid unnecessary change logs

CompositeProvider Design​

  • Avoid deep join chains
  • Prefer UNION over JOIN when possible
  • Filter early

Query Design​

  • Fewer free characteristics
  • Use filters instead of structures where possible
  • Avoid logic-heavy CKFs

6. BW/4HANA Modeling Best Practices (Consolidated)​

DOs

Follow LSA++ layers
Use Standard aDSO for harmonization
Use CompositeProviders for reporting
Keep queries lightweight
Design for reuse

DON'Ts

Don't recreate star schema thinking
Don't persist data just for joins
Don't duplicate business logic
Don't mix layers

7. BW/4HANA vs Classic BW Modeling​

AspectClassic BWBW/4HANA
Core ModelStar schemaLSA++
PersistenceHeavySelective
AggregatesMandatoryEliminated
ReuseLimitedStrong
PerformancePre-calculatedRuntime pushdown

8. Real-World Modeling Example​

ECC / S4
 ↓
Staging aDSO
 ↓
Standard aDSO (Sales Facts)
 ↓
CompositeProvider (Sales + Customer + Time)
 ↓
BW Queries

9. Interview-Grade Questions​

Q1. Why is star schema discouraged in BW/4HANA?​

Answer: Because BW/4HANA leverages HANA pushdown and layered architecture, making rigid star schemas and aggregates unnecessary and inflexible.

Q2. What is semantic layering?​

Answer: Semantic layering is the practice of separating technical data capture, business logic, and reporting logic into distinct layers to ensure reuse, performance, and maintainability.

10. Summary​

  • Star schema is replaced by LSA++
  • Semantic layering is mandatory
  • Reuse is a design goal
  • Performance is driven by pushdown
  • BW/4HANA modeling is architecture-first

11. What's Next?​

➡️ Module 9: Data Acquisition & Delta Management

Learning Tip

In BW/4HANA, good modeling beats clever tricks.