The dynamic replica placement problem with service levels in content delivery networks: a model and a simulated annealing heuristic

Dynamically assigning copies (replicas) of internet files to regionally placed servers is a crucial planning problem for content delivery network providers. Their goal is to minimize placement, storage and delivery costs while fulfilling service-level agreements. Although there is considerable work on simplified versions of this problem such as the static case, the dynamic problem with all three cost types and service-level constraints has yet not been considered in the literature. This paper provides a mixed integer programming (MIP) formulation for the problem as well as a simulated annealing metaheuristic. A computational study is employed to assess the simulated annealing heuristic and the MIP employing a state-of-the-art solver.