How-Rick-van-der-Lans-and-SuperNova-Data-Vault-Shaped-My-Approach-to-Data-Architecture

Applying Data Vault principles (without the dogma) to build model-driven engineering with a historized dual SCD2 core and flexible view-based delivery.

Summary

I take Rick van der Lans’s ideas — especially SuperNova, virtualization, and Data Vault principles — but skip the rigid HL&S (Hubs, Links, Satellites) structures.

Instead, I use a 3NF business data model, historized via dual SCD2, to get auditability, history, and business key orientation.

Views and virtualization are the first choice for delivery; physical materialization only when needed.

This gives me agility, clarity, and performance without complexity overhead.

Introduction

Over the past few years, I’ve been inspired by Rick van der Lans’s work: especially his SuperNova concept, his whitepapers, and his talks at conferences such as the Data Vault Alliance and DW & BI Summit. Those sessions helped me clarify what I believe a modern, agile data architecture should enable: delivering business value fast, retaining history and auditability, ensuring clarity, while avoiding unnecessary complexity.

At the same time, I came to see that implementing Data Vault literally (with separate Hubs, Links, Satellites) can bring more complexity and maintenance cost than necessary, depending on context.

I made a decision (in “Lessons from Data Vault: Principles Without the Dogma”) to borrow the principles — business key orientation, historization, separation of concerns — but implement them using a 3NF business model, historized via dual SCD2. This article explains how Rick’s work influenced that decision, and how those ideas map into my proposed architecture.

Rick’s Key Ideas That Resonated (Modified for My Context)

1. SuperNova & Virtualization
   Rick’s SuperNova idea emphasizes using views over a core model (or vault-like structure) to deliver business semantics without proliferating physical marts. Even though I didn’t use the vault literal structures, this view-first delivery layer remains central in my architecture.
2. Principles Over Patterns (Avoiding HL&S Dogma)
   In “Lessons from Data Vault: Principles Without the Dogma,” I articulate that the real value comes from principles: separation of concerns, business key orientation, historization, agility. Rick’s work gave me confidence that I didn’t need to mirror Hubs, Links, Satellites if I could keep these principles in place. Medium
3. Business Key Orientation & Auditability
   One constant theme in Rick’s work is that business keys are stable anchors for integration and consumption. Even though I’m using a 3NF model, business keys drive much of the schema design. History (audit, as-of timestamps, source metadata) is preserved via dual SCD2 forms.
4. Layering + View-Based Delivery
   Rick’s emphasis on layering (vault / business vault / views / exposure) strongly influenced my adoption of a view-based consumption layer: delivery through views, exposing business semantics, with physical materialization only when usage or performance demands.
5. Query Assistance & Performance Pragmatism
   While SuperNova and Data Vault include PIT tables, bridge tables, and query-assisting artifacts to mitigate performance challenges from many joins, I adopt similar patterns. Even if the core is 3NF + dual SCD2, stacking many views can degrade performance; so query assist layers, selective materialization, and monitoring usage become essential tools.

My Architecture: Dual SCD2 + View-First Delivery

Here’s how I have structured the architecture, combining those influences with my preference for a 3NF business model and dual SCD2 historization:

Press enter or click to view image in full size

How This Differs from Literal Data Vault Implementation

• No separate Hubs, Links, Satellites tables. Instead, the 3NF business model handles relationships and descriptive attributes.
• Use of dual SCD2 rather than satellites to store history. Every dimension or entity has both current and historical values.
• Less object count: fewer tables, fewer joins on retrieval paths, since not having separate link/hub/satellite artifacts reduces model complexity.
• More straightforward mapping between business schema and consumption views, since the business model is more directly human-understandable.

Challenges & Trade-Offs

• Some of the flexibility in “vault-style” architecture (parallel loading, loosely coupled sources) comes from the hub/link design. Without that separation, careful design is needed to avoid coupling.
• Joins may still increase in number, especially historically; need good performance practices.
• The dual SCD2 model must have well-defined surrogate keys, business keys, as-of timestamps; governance must ensure consistency.
• View stacking and virtualization may lead to latency/performance issues if not managed; selective materialization and query assist layers are essential.

Conclusion

Rick van der Lans, through SuperNova and related work, influenced me to build an architecture that holds agility, clarity, history, and performance close together. But I also realized that I didn’t need to follow every pattern (HL&S) literally to get the benefits.

By combining his principles with:

• a 3NF business data model,
• historized dual SCD2 instead of satellites,
• a view-first, virtualized delivery layer,
• and selective materializations + query assist where needed,

…I believe you can get most of the benefit of Data Vault + SuperNova with less complexity, quicker delivery, and easier consumption.