An approach to the architecture of distributed databases is presented, oriented around ex-tensive embedding of database primitives in the underlying distributed operating system. it is shown that this approach greatly reduces the effort required to provide high quality distributed database services substantial portions of this approach are implemented and are operational as extensions to the locus distributed Unix operating system. this paper describes the approach compares it with other work argues why it is attractive and what drawback are present, and lays out what has been done and what remains.