A method of optimizing the delivery of content over a content delivery network to client devices in accordance with optimization rules is provided. Requests for allocation and release of content delivery resources over a content delivery network are monitored, and the processing capabilities of client devices receiving content over the content delivery network are determined. Network resources are re-allocated based on optimization rules and the processing capabilities, and the client devices are directed to use re-allocated resources for reception of content. Apparatus and client devices for performing the method are also disclosed.