Discovering Bounds of Performance of Self-Organizing Content Delivery Systems

Self-organizing methods can efficiently search, route and replicate content in complex, dynamic networks. Furthermore, they make the assumption that decisions of the nodes of the networks rely only on local information and therefore the global optimum is not known. For evaluation purposes, however, it is important to compute the global optimum to serve as a theoretical bound. In this paper we define a formal model describing the problem of content placement and use an integer linear programming (ILP) based optimization method. With this method we discover the Quality of Service (QoS) bounds of a self-organizing content delivery system. We further demonstrate how to balance between run-time complexity and accuracy of the model by applying a use case.