SCARI: A Strategic Caching and Reservation Protocol for ICN
by Susmit Shannigrahi, Chengyu Fan, and Christos Papadopoulos
In the last decade, the Internet service model has shifted from sharing resources to distributing content. Certain applications such as large science data, CDNs, VoIP and multimedia applications transfer a significant amount of data over the Internet in a time-constrained manner that requires guaranteed in-network resources. The networking community often achieves this by creating point-to-point guaranteed bandwidth paths. However, the current resource reservation solutions are end-to-end and often initiated by users without the knowledge of the underlying network conditions. As a result, the data flowing through these reserved tunnels are not reusable, and in-network resources are not optimally utilized. On the contrary, Information-Centric Networking architectures such as NDN[16] has several properties that can facilitate resource reservations more intelligently. In this work, we present Strategic Caching And Reservation in ICN (SCARI), a protocol for reserving resources on ICN networks. Similar to RSVP[17] in IP networks, SCARI acts as a signaling mechanism before the actual data transfer. In this work, we focus explicitly on scientific dataflows and not on other types of traffic such as real-time audio/video. We show that SCARI reduces network resource consumption by aggregating reservations and strategically caching content in the network.