Abstract: The distribution of replica caches of information is dynamically optimized among the regions of a network, based upon the changing usage patterns. Initially, a single server in a first region stores the primary copy of information. A router in the first region receives service request messages from the plurality of users throughout network, directed to the server storing the primary copy. To determine if it is justified to deploy a replica cache, the router transmits a monitor request message to other routers in other regions of the network, requesting them to monitor all service requests in their respective regions, being directed to the primary copy in the first server. The first router then computes an optimum assignment of a replica copy of the information, for storage in at least one newly assigned server in the network.