No abstract

The power of Information-Centric Networking (ICN) architectures lies in their abstraction for communication-the request for named data. This abstraction promises that applications can choose to operate only in the information plane, agnostic to the mechanisms implemented in the connectivity plane. However, despite this powerful promise, the information and connectivity planes are presently coupled in today's incarnations of leading ICNs by a core architectural component, the forwarding strategy. Presently, this component is not sustainable: it implements both the information and connectivity mechanisms without specifying who should choose a forwarding strategy-an application developer or the network operator. In practice, application developers can specify a strategy only if they understand connectivity details, while network operators can assign strategies only if they understand application expectations. In this paper, we define the role of forwarding strategies, and we introduce Information-Centric Transport (ICT) as an abstraction for cleanly decoupling the information plane from the connectivity plane. We discuss how ICTs allow applications to operate in the information plane, concerned only with namespaces and trust identities, leaving network node operators free to deploy whatever strategy mechanisms make sense for the connectivity that they manage. To illustrate the ICT concept, we demonstrate ICT-Sync and ICT-Notify. We show how these ICTs 1) enable applications to operate regardless of connectivity details, 2) are designed to satisfy a predefined set of application requirements and are free from application-specifics, and 3) can be deployed by network operators where needed, without requiring any change to the application logic.

Linux software routers have many configuration options, most of which have received little attention from researchers. Over time, multiqueue NICs, NUMA architectures, and many changes to the kernel have altered the forwarding landscape. Here we investigate (i) allocation of NIC queues to processing cores, (ii) batch sizes at various layers, (iii) receive and transmit packet steering, and (iv) how FIB performance scales. Our experiments focus on forwarding minimum size packets, which maximize the packet processing stress for the kernel, at 10 Gbps. Our investigation uses Bayesian factor analysis, experimental design techniques, and kernel density estimation to robustly measure each factor's effect on the mean packet forwarding rate and mean RTT. This study does not seek bleeding edge performance. Rather, we elucidate key decisions for packet forwarding performance on a widely used platform. Our results show irqbalance is quite volatile, hyperthreading can provide over a 1 MPPS boost in forwarding performance, larger queue sizes do not improve forwarding but can add nontrivial latency (up to 30 ms), and both receive packet steering and low values of rx-usecs can induce receive livelock.