There is an increasing number of autonomous, portable, electronic devices that we use in everyday life, work and leisure, like notebook computers, personal digital assistants, palmtops, pocket pcs, mobile phones, digital cameras. As a result, a great amount of "mobile" data and services -necessary to individuals or communities of users- resides on these devices in a distributed, non-integrated manner. This data is not only heterogeneous but also carries context-dependent semantics (depending on user, location, community and other parameters). Furthermore, it demonstrates variable availability in time, depending on mobility parameters and network coverage, and can be accessed only locally, by using the device directly (depending on the type of the device). As a result, a virtual super-database with a dynamic schema is formed and a lot of new issues arise concerning its use. Traditional DB management, querying, caching, indexing techniques are non-applicable in this context and new paradigms are necessary. A lot of current projects deal with the issue of distributed computing environments and the problem of sharing, locating and accessing resources like CPU cycles and data storage in such a context. The DBGlobe project will attempt a data-centric approach, i.e. find innovative ways to dynamically describe, integrate, index, cache, search, retrieve, etc local data and services in such an environment in an optimal, uniform manner, transparently to the user, offering increased availability of the resources, scalability and fault-tolerance of the system.