A pragmatic architecture pattern for modern enterprise data platforms
Data platforms have evolved significantly over the past decade. Organizations are increasingly adopting Lakehouse architectures built around Bronze, Silver, and Gold layers, while Data Vault remains one of the most effective approaches for managing enterprise data integration, historical tracking, and auditability.
For many teams, this raises an important question:
Where does Data Vault belong in modern Medallion architecture?
After implementing Data Vault across multiple cloud-based data platforms, I've found that the most successful answer is rarely rooted in strict methodology. Instead, it comes from understanding the strengths of each architectural pattern and combining them in a way that serves both the platform and the business.
The result is a pragmatic approach that leverages the governance and historical capabilities of Data Vault while embracing the operational simplicity of modern Lakehouse architecture.
The Medallion architecture organizes data according to its level of refinement and intended purpose.
Data Vault addresses a different challenge.
Rather than organizing data by processing stage, Data Vault provides a framework for integrating enterprise data using business keys, relationships, historical tracking, and source lineage. It creates a scalable model capable of adapting as systems, processes, and business requirements change over time.
When viewed through this lens, the two approaches are highly complementary rather than mutually exclusive.
In many modern Lakehouse implementations, Data Vault serves as the primary integration model within the silver layer, creating a governed foundation between raw ingestion and business consumption.
|
Bronze |
Silver |
Gold |
|
Source-aligned raw retention, ingestion, replay, and reprocessing |
Data Vault integration using hubs, links, satellites, and selected consumption helpers |
Business-ready marts, semantic models, reporting datasets, and analytics products |
Traditional Data Vault implementations often distinguish between a Raw Vault and a Business Vault.
That pattern has proven successful for many organizations. However, modern Lakehouse platforms introduce capabilities that can change the equation.
With Medallion architecture, the bronze layer already fulfills many of the responsibilities historically assigned to a Raw Vault. It provides long-term retention of source data, supports replay and reprocessing scenarios, and preserves a consistent source of truth for downstream processing.
In some environments, maintaining an additional Raw Vault dedicated primarily to storing unmodified source data can introduce complexity without delivering proportional value.
A more streamlined approach can be to retain raw source data in Bronze while implementing Data Vault structures directly within Silver.
Under this model:
This approach preserves the core principles of Data Vault while reducing unnecessary duplication of responsibilities across the platform.
Within the Silver layer, Data Vault provides a clear separation of concerns:
Together, these structures create a scalable, auditable foundation for enterprise integration.
As new source systems are introduced, organizations can onboard data without extensive redesign. Existing business entities remain intact, historical context is preserved, and lineage remains traceable throughout the platform.
Perhaps most importantly, Data Vault enables organizations to manage change without overwriting history.
Instead of replacing prior values, changes are recorded over time, creating a complete record of how business information has evolved. This capability becomes increasingly valuable as governance requirements, regulatory obligations, and analytical complexity grow.
One area where implementation can depart from a strictly traditional Data Vault model is the use of bridge tables.
As Data Vault models expand, navigating large collections of hubs, links, and satellites can become challenging for downstream consumers. While the underlying architecture remains highly flexible and scalable, analysts and reporting teams frequently need a more approachable way to access integrated business concepts.
Bridge tables help address this complexity and several related challenges:
Rather than replacing Data Vault structures, bridge tables provide a curated layer that organizes frequently used relationships into business-friendly representations.
For example, a bridge table may consolidate multiple identifiers associated with the same person, simplify navigation of organizational hierarchies, or provide a unified view of a business entity assembled from multiple source systems. This allows consumers to work with familiar business concepts while preserving the lineage and traceability maintained within the underlying vault.
In our experience, bridge tables serve as an effective transition point between the highly normalized integration model and the business-oriented data products consumed by reporting, analytics, AI applications and downstream applications.
They allow the Data Vault layer to remain focused on integration while making enterprise data significantly easier to understand and use.
One lesson has become increasingly clear across modern data platform projects:
Architectural methodologies should support business outcomes, not become outcomes themselves.
Data Vault 2.0 provides an excellent framework for enterprise integration. At the same time, every organization operates within its own constraints, delivery timelines, support models, governance requirements, and technology landscape.
The most effective solutions often come from adapting proven patterns rather than implementing them dogmatically.
In several implementations, we intentionally prioritized maintainability, operational simplicity, and usability over strict adherence to methodology. Those decisions reduced complexity while preserving the capabilities that made Data Vault valuable in the first place:
The result was a platform that could evolve alongside the organization while remaining understandable and supportable for the teams responsible for operating it.
Data Vault and Medallion architecture are often discussed as separate architectural concepts, but they address fundamentally different challenges.
The Medallion architecture defines how data progresses through a platform. Data Vault defines how integrated enterprise data is modeled, historized, and governed.
When combined thoughtfully, they create a powerful foundation for modern data platforms.
For many organizations, that means using Bronze as the source-aligned retention layer, Data Vault as the governed integration model within Silver, and Gold as the layer where curated business data products are delivered to consumers.
The most successful data platforms are rarely the ones that follow a methodology most rigidly. They are the ones that apply proven architectural principles in ways that best serve the needs of the organization.
In the end, good architecture is not about perfection. It's about delivering a platform that is scalable, maintainable, trustworthy, and capable of adapting to whatever comes next.
At Spyglass, we help organizations design modern data platforms that balance governance, scalability, and delivery realities across Microsoft Fabric, Azure Databricks, Azure Data Services, and enterprise analytics ecosystems.