Data vault is a modern approach to design enterprise data warehouses. The two key benefits of data vault are data model extensibility and reproducibility of reporting results. Unfortunately, from a query and reporting point of view a data vault model is complex. Developing reports straight on a data vault-based data warehouse leads to very complex SQL statements that almost always lead to bad reporting performance. The reason is that in such a data warehouse the data is distributed over a large number of tables.
To solve these performance problems, many organizations have developed countless derived data stores. In addition, a multitude of ETL programs is developed to refresh these derived data stores periodically. Although such a solution solves the performance problems, it introduces several new ones. Valuable time must be spent on designing, optimizing, loading, and managing all these derived data stores. The existence of all these extra data stores diminishes the intended flexibility that organizations try to get by implementing Data Vault. Philosopher and statesman Francis Bacon would have said: “The remedy is worse than the disease.”
An alternative approach exists that avoids the performance problems of data vault and preserves its key benefits: reproducibility and data model extensibility. This new approach is based on data virtualization technology together with a data modeling technique called SuperNova. With this approach a data virtualization server replaces all the physical and derived data stores. So, no valuable time has to be spent on designing, optimizing, loading, and managing these derived data stores, and therefore reproducibility of reporting results and data model extensibility is retained. Also, because new derived data stores are completely virtual, they are very easy to create and maintain.
By combining data vault with data virtualization, the data stored in data vault-based data warehouses is easier to access and exploit, thus increasing its ROI. For more details on the SuperNova approach see this whitepaper.