Exploiting locality for data management in systems of limited bandwidth
The static and dynamic data management are investigated in computer systems where the computing nodes are connected by a relatively sparse network. In the static model, it is assume that the given application for which the rates of read and write accesses for all node-object pairs are known. The static placement of the objects to the nodes in the network is calculated and network congestion routing is minimized. In the dynamic model, no knowledge about the access pattern is assumed. Dynamic caching strategies are developed to minimize the congestion on trees, meshes, and clustered networks.