Arnon Sturm

In recent years, new data characteristics led to the development of new database management systems named NoSQL. As opposed to the mature Relational databases, design methods for the new databases receive little attention and mainly consider the data-related requirements. In this paper, we present methods for designing two types of NoSQL databases – Document and Graph databases – that consider not only the data-related but also functional-related requirements. We empirically evaluate the methods and apply the design methods to two leading database systems. We found that the databases designed with the consideration of functional requirements perform better, with respect to time of execution and database I/O operations, compared to when designed without considering them.