Video on demand (VoD) systems currently use content delivery networks (CDN) to feed content to users, whose performance and effectiveness depends on the architecture, the number and the geographical location of CDN nodes deployed by CDN providers or ISP(s) itself. Content-Centric Networking (CCN), with the benefits of caching and sharing content by every node in the network, suggests an alternative: a collaborative caching system exploiting the maximum capacity of infrastructure for the high performance of video delivery services. However, a CCN-based architecture to support efficient VoD delivery raises important questions about the optimal routing and caching strategies with constraints on the architecture and capacities of the system. We investigate models and algorithms for addressing these optimization problems. We study different solutions for the routing and caching optimization problems and compare the solutions produced with the optimal solution under various assumptions. Our numerical results show the influence of throwing caching at the problem in different locations, on the system performance and its related cost.