Information-centric networking (ICN) has generated a tremendous amount of interest in recent years.  ICN promises benefits for users and service providers along several dimensions, including performance, security, and mobility. These benefits, however, come at a non-trivial cost as many ICN proposals envision a significant upgrade to the Internet infrastructure: routers need to act as content caches and support nearest-replica routing.  In this talk we explore whether the additional complexity is justified and whether these benefits can be realized in an incrementally deployable fashion.  We begin by performing trace-driven simulations to analyze the quantitative benefits attributed to ICN (e.g., lower latency and lower congestion). Somewhat surprisingly, we find that pervasive caching and nearest-replica routing are not fundamentally necessary — most of the performance benefits can be achieved with edge caching architectures. We also discuss how the qualitative benefits of ICN (e.g., security and mobility) can be achieved without any changes to the network.  Building on these insights, we present a proof-of-concept design of an  incrementally deployable ICN architecture. 

Joint work with Seyed Fayazbakhs, Yin Lin, Amin Tootoonchian, Ali Ghodsi, Teemu Koponen, K.C. Ng, Vyas Sekar, and Scott Shenker.