A forwarding plane for information-centric internetworking Christian Esteve Rothenberg, Maurício Ferreira Magalhães (Orientador) Departamento de Engenharia de Computação e Automação Industrial (DCA) Faculdade de Engenharia Elétrica e de Computação (FEEC) Universidade Estadual de Campinas (Unicamp) Caixa Postal 6101, CEP 13083-970 – Campinas, SP, Brasil {chesteve, mauricio}@dca.fee.unicamp.br Abstract – This paper presents several areas of ongoing doctoral research towards the design and evaluation of network functions to enable an information-centric forwarding plane. With the hypothesis of a future Internet around large scale information-object interconnection, classical host-centric routing approaches cannot work efficiently and new communications schemes should be developed. Target components and network support functions include an internetworking framework based on hash-based data structures to enable efficient multicast transport services. Research issues include understanding the implications of network coding and opportunistic caching in an information-centric multicast-capable forwarding fabric. Keywords – Internet architecture, forwarding, source routing, multicast, Bloom filter, network coding, caching 1. Introduction The Internet has shifted from being a simple host connectivity infrastructure to a platform enabling massive content production and delivery, transforming the way information is generated and consumed. From its original design, the Internet carries datagrams inserted by sending hosts in a best effort manner, agnostic to the semantics and purpose of the data transport. Many efforts have been undertaken to get around Internet host-centric limitations in terms of mobility, security, address space exhaustion, routing and content delivery efficiency. Continuously patching the TCP/IP suit with ad- hoc protocol extensions and overlay solutions (e.g., NAT-aware protocols, CDN, DPIs, etc.) is regarded as a complex and costly solution for the long term. Recent research to circumvent current Internet limitations advocates the investigation of a completely new architecture (clean-slate). From a research perspective, “clean-slate” design does not presume “clean-slate” deployment and aims at innovation through questioning fundamentals. In this context, a key question is to what extent a new paradigm for the future network is really necessary, e.g., as packet switching was to circuit switching in the 70‟s. The reasoning is based on the large scale use of the Internet for dissemination of named pieces of data [4]. Tons of connected devices are generating and consuming content, without caring about the actual data source as long as integrity and authenticity are assured. Such an internetworking design principle has been baptized as “content- centric” or “information-oriented” networking, and aims at shaping the future Internet to interconnect information at large rather than to provide plane device connectivity. A shift in the orientation of network architecture design implies rethinking many fundamentals, starting by having a new identifier space for information objects and individual data items is required and enabling thereby more expressive communication patterns (e.g., publish/subscribe), higher network efficiency (e.g., native multicast, opportunistic caching, network coding) and increased resilience (e.g., security, data replication, multi-path, lateral error correction). 2. Background As a specific example, the EU FP7 PSIRP project [7] has a goal of building a pure publish/subscribe-based network, where the architecture uses pieces of data as the first-class citizens. With the publish operation, an endpoint can indicate that it wants to associate a document or a one-way channel with the given (possibly randomly looking) identifier. With the subscribe, an endpoint can signal its desire to get (read- only) access to the named document or channel. Based on the subscriptions, the network is responsible for delivering the document or any data appearing on the channel, to all the subscribers. As typical to pub/sub systems, multicast is the natural mode of communication. The present research can be regarded as an evolutionary step of the Forwarding and more network functions to enable the RTFM architecture [6] that consists of the following recursive functional building blocks: