The effect of video caching on network resource planning — a real-case study

Traffic Engineering is one of the building blocks for a correct network planning. Internet Service Providers are always trying to fulfill the user Quality of Experience (QoE). However, each technological advance brings new services to the user, with new challenges to be solved to maintain the QoE. An example of this feedback system is the increased access speed in the user's premises. The increased bandwidth allows the Internet-based video services, but the increased load also requires new techniques to prevent congestion in the core network. A quite common system to lower the network load and increase the system availability is is represented by Content Delivery Networks (CDNs). However, the optimization of the CDN servers optimization is still an open problem. In this paper we will analyze the performance of a deployed Content Delivery Network caching system. We will derive its performance metrics and we will show how a properly chosen metric can indicate if the cache is underutilized. We will also demonstrate that some `logical' assumptions are not true in the vast majority of cases, and that moving the cache close to the users is negatively impacting the network performances in most cases.