Evaluation of Replica Placement and Retrieval Algorithms in Self-Organizing CDNs

Content distribution networks (CDN) have been increasingly used as the scalable solution to deliver high-quality multimedia content. However, despite the promising concept, current self-organizing content distribution networks (using realtime replica placement) lack placement and retrieval algorithms that are both intelligent and scalable. In this paper we present an architecture and relating algorithms for such a scalable, selfoptimizing content delivery service. Contrary to distributed and centralized CDN architectures, we propose a hybrid approach. In order to tackle specific problems such as congested network parts and the occurrence of flash crowds, network monitoring and multi-source traffic engineering are combined.