The increased availability of meta-data in Web 2.0 (as opposed to traditional Web) can and should be exploited to make such techniques more effective. Caching should improve the performance and scalability of multimedia service streaming (e.g., YouTube). In this contribution, we introduce new directions and considerations in the analysis of caching popular content in the Web which allows us to gain insight on deriving more informative indications for quality of service development. We provide a dynamic model for the impact of popularity on the access speed due to caching policies of a service provider. More specifically, we assume that caches are spatially deployed as a Poisson distribution and that users are distributed over the geographical area in a Poissonian manner. Our model is formulated as epidemic type process of file dissemination. We then study the transient behavior of caches where information is replicated and disseminated according to an epidemic type dynamics based on the popularity of the content. Simulation results show that the proposed scheme provides significant improvement in terms of the system throughput.