Distance-Dependent Distributed Directories

Distributed systems often make use of directory servers enabling the storage and retrieval of global information in the network. In many practical situations, the data are directly related to particular sites of the network, and therefore the desired retrieval characteristics are dependent on the network topology; for example, priority is given to having high accessibility to data generated in nearby locations. This paper defines the concept of distance-dependent directories, and then proposes appropriate accessibility measures and presents strategies for constructing such directories. The construction methods are based on efficient solutions to a new type of graph-covering problem. As an application, it is shown how to use a distance-dependent directory to implement a name-server component for a routing scheme.