Mixed Modeling DWH

Mixed Modeling Approach in AnalyticsCreator

The mixed modeling approach combines elements of different data modeling strategies—most commonly Kimball (dimensional modeling) and Data Vault—to meet modern enterprise data warehouse needs. It leverages the strengths of both approaches to optimize performance, data governance, historical tracking, and agility.

When and Why to Use a Mixed Modeling Approach

Enterprises are increasingly dealing with both structured and semi-structured data, frequent business rule changes, and the need for both auditability and performance. Relying on a single modeling paradigm is often not sufficient.

• Use Kimball-style models in the data presentation layer to support fast query performance and ease of use for BI tools.
• Use Data Vault in the raw data layer to handle changing business logic, full historization, and traceability.
• Mix both when you need governance, auditability, and flexibility without sacrificing performance and usability.

How the Mixed Model Works in AnalyticsCreator

AnalyticsCreator supports a mixed modeling approach by allowing users to define the logical and physical layers separately using metadata. This flexibility is built into the platform’s model-driven architecture.

• Model core business entities using Data Vault (Hubs, Links, Satellites) to ensure historization and auditability.
• Expose business-friendly Kimball-style dimensions and facts from the raw vault or stage views.
• Use model variants in AnalyticsCreator to define parallel data marts or reporting models on top of the same raw layer.
• Deploy these models directly into Fabric SQL for governed data storage and OneLake Delta Tables for consumption.

Benefits of the Mixed Modeling Approach

Limitations and Considerations

• Initial setup of both modeling layers requires strategic planning and governance.
• Data Vault structures may be less intuitive for business users if directly exposed.
• Requires a platform like AnalyticsCreator to manage model complexity and deployment consistency.

Mixed Modeling in Fabric with AnalyticsCreator

AnalyticsCreator simplifies the deployment of mixed modeling architectures into Microsoft Fabric:

• Fabric SQL Databases: Hosts the raw vault, stage layer, and dimensional models using the metadata-generated schema.
• Azure Data Factory Pipelines: Automatically generated to handle data ingestion and ETL into the appropriate layers.
• OneLake Delta Tables: Serve as consumption endpoints for Power BI and other tools, supporting both real-time and batch scenarios.

By combining these technologies with a mixed modeling strategy, you gain a balance of governance, performance, and adaptability at scale.

Use Case Example

A global retail company implemented a mixed model in AnalyticsCreator to meet audit requirements while supporting self-service BI. They modeled transactional data with Data Vault to preserve history and compliance. On top of the raw vault, they generated conformed dimensions and facts for finance and supply chain reporting in Power BI.

Thanks to AnalyticsCreator’s metadata engine, they deployed both models into Microsoft Fabric with one-click publishing, enabling a modern, governed, and flexible analytics platform.

Key Takeaway

The mixed modeling approach in AnalyticsCreator enables you to build auditable, high-performing, and scalable data warehouses on Microsoft Fabric. By blending the strengths of Kimball and Data Vault, and automating the deployment using metadata, organizations reduce risk and speed up delivery.

Next Steps

Want to see how a mixed model would look in your Fabric environment? Book a technical session with our team to explore your use case.