Caching in networks

We present a general framework for the development of on-line 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, fat-trees, hypercubic networks, and complete networks. For example, our framework yields an $O(d cdot log n)$-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 modules due to remote accesses. Previous work on data management in networks either neglects memory capacity constraints or minimizes only the total communication load, i.e., the sum of the communication load over all resources, which may produce congestion as some of the links become bottlenecks.