Due to the current technological and economical context, there is an increa
sing need for cooperating information systems based on federated databases.
Though the technical issues of these architectures have been studied for l
ong, the way to build them has not triggered as much effort. This paper des
cribes a general architecture, a methodology and a CASE environment intende
d to address the problem of providing users and programmers with an abstrac
t interface to independent heterogeneous and distributed databases. The arc
hitecture comprises a hierarchy of mediators and a repository that dynamica
lly transform actual data into a virtual homogeneous database and allow cli
ent applications to query it. The InterDB approach provides a complete meth
odology to define this architecture, including schema recovery through reve
rse engineering, database integration and mapping building. The methodology
is supported by the DB-MAIN CASE tool that helps developers generate the m
ediators and their repository.