Caching in Networks

d Abstract) Friedhelm Meyer auf der Heide   Berthold V¨ocking† Matthias Westermann   We present a general framework for the development of online algorithms for data management in networks with limited memory capacities. These algorithms dynamically create and delete copies of shared data objects that can be read and written by the nodes in the network. Our algorithms aim to minimize the congestion, i.e., the maximum communication load over all network resources, so that that none of these resources become a communication bottleneck. We give several examples of networks for which our framework yields efficient algorithms, including meshes, fattrees, hypercubic networks, and complete networks. For example, our framework yields an O ¡ d ¢ logn£ -competitive caching algorithm for d-dimensional meshes of size n with respect to the congestion on the communication links, and an O ¡ 1£ -competitive algorithms for complete networks with respect to the congestion at the memory modul...

• Type: Conference
• Year: 1999
• Where: GI
• Authors: Matthias Westermann
• Keywords: Algorithms, Communication Load, Complete Networks, GI 1999, Theoretical Computer Science