The paper describes an architecture designed to allow distributed hierarchies to be modelled and controlled. The use of node and network link capabilities and file importance gradings is used to model and control the resultant highly distributed server hierarchy. These gradings together are used to form file interests which, in the case of file replication, determine how files and updates are propagated throughout the distributed hierarchy, and how often updates are performed. These gradings encapsulate most of the behaviour required in a distributed storage server architecture. The paper discusses the load balancing techniques used in the system, and the distribution of the hierarchy and its formation. The use of the previously described capabilities in the modelling and control of the hierarchy is then discussed, followed by a description of the design of the individual storage nodes.