XNorthwind: Grammar-driven synthesis of large datasets for DB applications

Relational databases often come with sample databases. One known example is the Northwind database, often used as data repository for software testing and development purposes. The Northwind database includes hypothetical records of customers, companies, products, employee and so on. The number of records in the Northwind is however considered inadequate for large applications, where a developer or user may need a lot more, possibly, millions of records. In this paper, we have used a Context-free Grammar in describing the rules for the synthesis of exponentially many hypothetical datasets that are similar to the Northwind database. We referred to the resulting database as XNorthwind (Extended Northwind). The new grammar was implemented, resulting in thousands of unique data values across the eight different Northwind Data Tables. These datasets will find applications in training and development environments. A survey of 112 participants' perceptions showed that 94.6% agreed that the XNorthwind can be useful.