A Content Delivery Network (CDN) aims to provide efficient movement of massive digital content (multimedia and files) across the Internet. This is achieved by putting the content in servers closer to the costumer. Video-On-Demand service is an application of CDN where videos have to be located strategically to avoid network congestion and servers saturation. Therefore, the problem of optimal placement of videos arises. This problem has a block diagonal structure with linking constraints on links and servers capacities. In this project, we solve huge instances of a video placement problem over three real network topologies with a specialized interior point solver named BlockIP. The evaluated instances range from 7 to 300 millions of variables and the difficulty of the instances depends on the size of servers, links bandwidth and network topology. Our results: 1) verified characteristics of BlockIP like regularization and the intensive computation in the last iterations and 2) showed that BlockIP found optimal solution in all the evaluated instances with a good optimality gap. On the contrary, state-of-art CPLEX cannot reach an optimal, feasible solution in some difficult instances and needs almost twice the memory of BlockIP. However, CPLEX solved most of feasible instances at least twice faster than BlockIP