What is Data Vault 2.0?
Data Vault 2.0 describes a design paradigm for data delivery. A key advantage of Data Vault 2.0 is its ability to respond very flexibly to changes in business processes without fundamentally modifying underlying structures. To achieve this, Data Vault 2.0, in contrast to a traditional data warehouse, is not the defined, aggregated truth of all data but rather the neutral provision of all available data. Data aggregation takes place on top of it, for example in data marts.
At its core, Data Vault 2.0 consists of 3 elements:
- Hubs, for the unique definition of a business object
- Links, to represent interactions/transactions between business objects
- Satellites, for the descriptive content of hubs or links Optionally, reference tables can be used for static data, e.g. currency information or country codes.
Due to this separation, each satellite can be loaded independently. This enables high levels of parallelization or adjustments tailored to the respective data source. Furthermore, the integrity regarding time and source of each satellite can be ensured through corresponding metadata. A hub represents only a unique, immutable identification of an element, e.g. a specific customer, a product or an order.
The descriptive content of an element is stored in one or more satellites. For example, a customer hub may have multiple satellites: one containing contact person data, one with delivery addresses and another with billing addresses.
Even if, technically speaking, delivery and billing addresses might be identical, they are stored independently in their respective satellites, as they may change independently and it is the responsibility of each satellite to provide the data valid at a specific point in time.
Links represent direct relationships between hubs, and their satellites provide the details of the transaction. In the given example, a link “order process” could connect order, customer and product, while a satellite of the link would store transaction-specific details.