This paper presents a methodology for using a generic data-driven spatial access method as a communication mechanism for vario-scale data in a server-client setting. As a complete data set is often quite large, it is managed at the server side and supporting different scale levels is important. We show that a generic R-tree like grouping method, commonly used for efficiently organizing and retrieving data from a database, can be used in a networked architecture and that it allows off-loading processing tasks from a server to a client. This helps in making web services more scalable. The method supports efficient retrieval of partial data by a client and makes it possible to reuse data by means of caching. This can make communication more efficient.