Skip to main content

CryspIQ® vs Data Vault 2.0

Data Vault solves getting data in; CryspIQ® solves getting answers out. Data Vault 2.0 is an ingestion and history pattern — hubs, links and satellites — designed for auditable, long-term storage of data arriving from many operational systems. It is deliberately not a consumption model: you still build star schemas downstream for anyone to query. CryspIQ®'s decomposed store is the consumption model, which removes the second modelling exercise rather than optimising it.

What Data Vault is​

Data Vault separates the things a business talks about (hubs, keyed by business keys), the relationships between them (links) and everything descriptive or time-varying about them (satellites). Loads are insert-only and non-destructive, so the warehouse keeps a complete, auditable record of what each source said and when it said it.

It scales well, it parallelises well, and in regulated environments the audit properties are genuinely valuable.

Where the two actually differ​

How many models you build. A Data Vault estate has two: the vault itself, and the marts built on top so people can query it. Both need designing, building and maintaining, and the second one reintroduces exactly the definition-drift problem the vault was supposed to prevent — because each mart re-decides what the numbers mean. CryspIQ® has one model, and the business definition is carried on the fact type rather than reapplied per consumer.

Business keys are still your problem. A hub is only as good as the business key chosen for it, which means Data Vault asks the same question every master data programme asks: what is a customer, and which source is authoritative? CryspIQ® answers it structurally — a single instance per party, enforced by the schema, not by key-selection discipline maintained across every load.

What a change costs. Both methods are built to absorb new sources, and both do. The difference shows up further downstream: in Data Vault a change that alters meaning still has to be pushed through the marts that depend on it, which is why change impact stays high. In CryspIQ® the consumption model is the same model, so there is nothing downstream to push it through.

Lineage. Data Vault gives excellent load lineage — you can always see what arrived and when. Tracing a number in a report back through the marts to the vault to the source is a separate exercise. CryspIQ®'s link key does that in one hop.

Who has to be in the room. Data Vault modelling is a specialist skill, and estates typically depend on people who have done it before. CryspIQ® is designed so that mapping a source is a review task for someone who understands the business, rather than a modelling task for someone who understands the method.

Where Data Vault is stronger​

Auditability. If your requirement is to prove exactly what every source system asserted at every point in time, non-destructively, Data Vault is purpose-built for it and CryspIQ® is not making that its central claim.

It also imposes no opinion about what your facts mean. Data Vault will model anything you can define a business key for, where CryspIQ® asks your data to fit seven fact types and a fixed set of dimensions.

Using both​

Where a vault already exists and is trusted as the historical record, CryspIQ® can map from it in the same way it maps from a raw or staging layer — the vault keeps doing what it is good at, and the marts built to make it queryable become the part that is no longer needed. See co-existence.