In this paper, an architecture and an implementation of a multipurpose relational database server are proposed. This architecture enables classical queries to be executed, deductions to be made, and data mining operations to be performed on fuzzy or classical data. The proposal of this integration is to combine several ways of querying different types of data. In order to achieve this, a combination of existing metaknowledge bases and new data catalog elements is presented. We also introduce a language for handling all these data coherently and uniformly on the basis of classical SQL sentences.