Do We Need Replica Placement Algorithms in Content Delivery Networks?

Numerous replica placement algorithms have been proposed in the literature for use in content delivery networks. However, little has been done to compare the various placement algorithms against each other and against caching. This paper debates whether we need replica placement algorithms in content delivery networks or not. The paper uses extensive evaluation of algorithms, systems and web workloads to answer this question. We conclude that a simple delayed-LRU caching scheme outperforms, or at least performs as well as the best replica placement algorithms. As LRU caching is less complex than most replica placement algorithms, caching is clearly the preferred option. However, we believe that replica placement algorithms will be necessary once properties such as availability, reliability, performance and bounded update propagation will have to be guaranteed by content delivery networks.