Optimization of cache expiration dates in content networks

Abstract

One of the fundamental decisions in content networks is how the information about the existing contents is deployed and accessed. In particular, in many content network architectures, there are access nodes which cache information about the contents location, in order to ensure a quick answer to queries formulated from the peer nodes. In this work we present a simplified model of the costs and restrictions associated with these cache expiration dates in a content network, which regulate the proportion of queries which will be answered on the basis of cached information, vs. those which will give rise to additional searches in the network backbone. This model gives rise to a mathematical programming formulation which can be useful to determine the optimal cache expiration dates in order to maximize the total information discovered, while respecting the operational constraints of the network. We apply the model to DNS data to obtain some insights about the behavior of the model and the optimization procedures.