Data Product Architecture

Bronze Zone — The Landing Zone

The Bronze Zone is the landing area for raw, untransformed application data before it is loaded into the Data Vault.

Its role is intentionally narrow:

• capture data safely

• preserve it exactly as it arrived

• enable replay or reprocessing when needed

Silver Zone — The Enterprise Source of Truth 

The Silver Zone is where data becomes enterprise truth.

This is where the Data Vault sits - a temporal, immutable Enterprise Data Warehouse that separates how data is produced from how the business understands it.

Data Vault does this using a simple, modular structure:

• Data Vault Spine - defines business keys (Hubs) and relationships (Links) shared by the Raw and Business Vaults

• Raw Vault - records temporal context (Satellites) for each source system 

• Business Vault - applies business rules to create curated, business-aligned temporal context (Satellites)

The Business Vault reflects enterprise meaning rather than application‑specific design. This keeps it closely aligned to the Business Concept Model and Enterprise Data Model and, with the right patterns, allows Hubs, Links and Satellites to be derived directly from them.

Data Vault allows the model to evolve while preserving full history and lineage - even when upstream data is imperfect. That imperfection is expected. No organisation has perfect source systems, and waiting for “perfect” data before ingestion only slows progress. Data quality should ideally be fixed at source, with improvements naturally flowing downstream.

It’s also important to be clear about what the Silver Zone is not.

It is not a consumption layer. It is not optimised for analytics, dashboards nor is it a data product - and it is rarely queried directly. Instead, Gold Zone data products are built from the Business Vault, which provides the most stable, business‑oriented view of data.

This design significantly reduces the impact of system change. When source applications are replaced, data products remain stable because they depend on enterprise truth - not system‑specific structures.

And because the implemented Data Vault model is bi‑temporal and immutable, data products can always be recreated accurately for any point in time, without maintaining their own complex history.

Gold Zone — Use Case Specific Data Products

The Gold Zone is where data finally turns into business value.

Data products are built using the application data model type best suited to the use case - relational, dimensional, wide‑table, graph, time‑series or others.

Each data product:

• aligns to a specific business domain or value stream

• uses business‑friendly semantics

• is optimised for its intended consumption pattern

Because full history is preserved in the Silver Zone, Gold Zone products retain only the history they actually need. This leads to lower storage costs, better performance, and simpler products that are easier to rebuild and evolve.

Data products are shared only when a single product genuinely serves multiple use cases - not as a compromise that satisfies none.

Data Domain Aligned Data Vault

The Data Vault is organised into domain‑aligned subsets, each owned by a data domain owner who is accountable for both structure and meaning.

Domain owners are accountable for:

• defining domain Hubs, Links, and Business Vault Satellites

• agreeing business rules and reference data

• ensuring end‑to‑end lineage exists for downstream consumers

• guaranteeing data remains bi‑temporal and immutable

Where data spans multiple domains, domain data architects collaborate to model the required Links and Satellites.

This operating model balances strong enterprise governance with domain autonomy - preserving consistency without creating a central bottleneck. 

Producer Data Mesh — Publishing the Truth

Producers are the teams that build and operate source systems - and they are the experts in the data those systems create. That makes them best placed to publish data into the Data Vault.

Producer teams work closely with data domain owners to understand which Hubs, Links and Satellites their data populates. They are responsible for loading Raw Vault Satellites, capturing data values exactly as they were produced. Where a producer is the sole source for records in a Business Vault Satellite, the domain owner may authorise them to load it directly.

When new attributes appear, they are loaded immediately into the Raw Vault so nothing is lost. Producers work with the domain owner so those changes are then reflected in the Business Vault.

Producers are accountable for:

• publishing immutable data

• correcting errors using bi‑temporal records, not overwrites

• completing lineage for the data they load

Depending on the ingestion pattern, the Bronze Zone - and in some cases even the Raw Vault - may be bypassed, for example when consuming pre‑curated event data.

Consumer Data Mesh — Creating Business Value

Consumers are experts in how data is used - not necessarily in how it is created.

In this architecture, they don’t need deep knowledge of source systems. All consumption happens from the Business Vault, where data is already cleansed, conformed and curated.

Consumers focus on:

• creating data products aligned to business outcomes

• choosing the right type of application data model for each use case

• delivering insights quickly

Consumers are billed for the compute and storage they use, encouraging them to retain only the history their use cases genuinely require.

They are accountable for being able to recreate their data products at any point in time, relying on the immutable, bi‑temporal Data Vault as the source of truth.

The result is a portfolio of data products that are lightweight, disposable and able to scale with demand. 

Data Platform

Infrastructure and tools for tenants to manage, process, store, govern and deliver data products

Data Fabric

Tooling that provides unified data management capabilities across distributed data platforms, simplifying access and control for users