Application-Mimes - An Approach for Quantitative Comparison of SQL - and NoSQL-databases

Due to the rise of NoSQL systems over the last years, the world of commercially applicable database systems has become much larger and heterogeneous than it was ever before. But the opportunities that are associated with the upcoming systems have also introduced a new decision problem that occurs in the information system design process. Once, benchmarking has helped to identify the proper database product among the de facto standard SQL systems. Nowadays, functional and non-functional properties of database systems and their implication on application development are so divergent that not all systems that come into account for realisation of a specific application can be covered by the same benchmark. In this paper we present an approach for experimental comparative information system evaluation that allows for well-grounded selection among diverging database systems. It is based on the concept of so-called application-mimes, i.e. functionally restricted implementations that focus exclusively on the information systems’ interaction with data management and try to mimic the target systems behaviour in this respect as realistic as possible.