Pervasive information systems give an overview of what digital environments should look like in the future. From a data-centric point of view, traditional databases have to be used alongside with non-conventional data sources like data streams, services and events to deal with new properties of such information systems including dynamicity, autonomy and decentralization. In this context, the definition of continuous queries combining standard relations, data streams and services in a declarative language extending SQL is clearly an ambitious and motivating goal. Those continuous queries could express an event management functionality (e.g. event filtering, event composition), associate events with data from legacy systems, and perform cost-based optimal calls to services. In this paper, we define virtual tables with binding patterns to represent services of the pervasive environment. By the way, relations, data streams and virtual tables can be homogeneously queried using a SQL-like language, on top of which query optimization can be performed. We also introduce a new clause defining the optimizing criteria to dynamically choose the best way to handle each event. A prototype on top of STREAM, a DBMS devoted to data streams, has been devised on which first experiments have been carried out on synthetic data.