A comparative study of Content-Centric and Content-Distribution Networks: Performance and bounds

Abstract

The Content-Centric Networking paradigm aims at improving the Quality of Service of the Internet by providing innovative features to better handle digital content distribution. A major step towards the success of this novel paradigm is to analyze and compare its performance with respect to the most popular ways in which content is disseminated in today's IP Internet. In this paper we give clear answers to this critical issue by proposing a methodology to assess how the innovative design of Content-Centric Networking behaves as opposed to the solution proposed by Content-Distribution Networks. We develop a novel optimization model to study the performance bounds of a Content-Centric Network, by addressing the joint object placement and routing problem. We further introduce three comparative models that well describe 1) a Content-Distribution Network, 2) a traditional IP-based network, and 3) a Content-Centric Network whose caches are pre-populated with given contents. To the best of our knowledge, our proposal is the first that studies the performance bounds of Content-Centric Networks by means of an optimization model. Finally, we discuss the numerical results showing the performance bounds of this revolutionary paradigm. We discover that: 1) a Content-Centric Network with small caches can provide significant performance gains compared to a traditional IP-based network; 2) for large amounts of caching storage, the benefits of using sophisticated cache replacement policies are dramatically reduced and 3) in some scenarios, a Content-Distribution Network with few replica servers can perform better than a Content-Centric Network, even when the total amount of available caching storage is exactly the same.