The challenge of peer-to-peer computing goes beyond simple file sharing. In the DBGlobe project, we view the multitude of peers carrying data and services as a superdatabase. Our goal is to develop a data management system for modeling, indexing and querying data hosted by such massively distributed, autonomous and possibly mobile peers. We employ a service-oriented approach, in that data are encapsulated in services. Direct querying of data is also supported by an XML-based query language. In this paper, we present our research results along the following topics: (a) infrastructure support, including mobile peers and the creation of context-dependent communities, (b) metadata management for services and peers, including locationdependent data, (c) filters for efficiently routing path queries on hierarchical data, and (d) querying using the AXML language that incorporates service calls inside XML documents. Keywords Peer-to-peer computing, peer-to-peer databases, global computing, ser...