Enabling Name-based Packet Forwarding over Flat ID Network Architectures

A BF-BASEDANALYTICALMODEL : DETAILS At its core, the analytical model of BF-based forwarding calculates the probability of forwarding decisions within a router, assuming Longest Prefix Matching (LPM) semantics. E.g. say a router has 3 interfaces 0, 1 and 2 each associated forwarding entries with sizes 1 or 2. Themodel calculates the probability of events such as ‘having a packet forwarded over interface 1 due to a match of size 1’. To better illustrate how the core of themodel works, we introduce a running example, depicted in Fig. 1. The main characteristics of the example are: • A topology with 5 routers: R0 to R4 • Each router holds forwarding entries of size 1 or 2 • Each router is the gateway to a lower-level access network which advertises1millionprefixes (representedascaches in thediagram). As such, each router has 1 million forwarding entries associated with the local interface (interface 0), and 4 million entries distributed over the remaining interfaces. • Following from the above, the forwarding table at each router holds a total of 5 million entries • The flow of the packet starts at router R0 • The requested content is held by router R2 only. Other routers do not hold the requested content. Let’s assume R2 announces that content using a prefix of size 1. • The shortest paths from R0 to any other router are represented by green dotted lines Forwarding decisions depend on the forwarding events which can occur at a BF-based router. In section A.1 we briefly explain what forwarding events are. The subsequent sections explain the steps involved in the calculation of forwarding event probabilities: (1) largest match probabilities per interface (section A.2); and (2) the actual calculation of the forwarding event probability (section A.3). Finally, we explain how to calculate a set of auxiliary probability R0