What is Data Vault modeling?

Data Vault is a data modeling pattern built around three table types: hubs (unique business keys, like a customer ID), links (relationships between hubs, like a customer placing an order), and satellites (the descriptive attributes and their history, attached to a hub or link). It's designed to absorb new sources and changing structure without redesigning the model.

How it works

A hub table holds just the business key and when it was first seen, nothing else. A link table records that two or more hubs are related. A satellite table holds the actual descriptive data, customer name, order status, and every version of it over time, tied back to a hub or link. Adding a new source usually means adding a new satellite, not restructuring existing tables, which is the pattern's main selling point for large, multi-source environments with sources that change shape.

Why it matters

Enterprises with dozens of source systems, each changing independently, need a model that doesn't break every time one source adds a column or a business rule shifts. Data Vault's separation of identity (hubs), relationships (links), and history (satellites) means a schema change in one source usually touches one satellite table, not the whole model. The tradeoff is more tables and more joins than a star schema, which is why Data Vault is often paired with a star schema built on top of it for reporting.

Where this stands with OptimaFlo today

OptimaFlo does not automatically build a Data Vault structure. The Ready (gold) layer is LLM-generated SQL that you review and approve before it runs, on top of the Clean layer, and that SQL could implement hubs, links, and satellites if you write it that way. Nothing in the platform detects a new source and generates a hub or satellite for it automatically today. If your team relies on Data Vault modeling, treat it as a pattern you build into your Gold SQL, not a capability OptimaFlo ships out of the box.

Related terms

• Star schema: The reporting-friendly model teams often build on top of a Data Vault for BI queries.
• Medallion architecture: The Raw, Clean, Ready layers where a Data Vault, if you build one, would live in Ready.