The Name-Based Routing Protocol (NBRP) performs routing by name with a structure similar to BGP [15]. Just as BGP distributes address prefix reachability information among autonomous systems, NBRP distributes name suffix reachability to content routers. Like BGP, NBRP is a distance-vector routing algorithm with path information; an NBRP routing advertisement contains the path of content routers toward a content server.
At its most basic, a BGP routing advertisement consists of an address range, a next-hop router address, and a list of the autonomous system (AS) numbers through which the advertised route will direct traffic. For example, an advertisement for Stanford's IP address range might specify 171.64/255.192 as the range, 192.41.177.8 as the next hop router, and 7170, 1 as the AS-path.
As shown in figure 4, a name-based routing advertisement contains essentially the same information. The advertised content is named example.com, the next hop toward that content is the address of the content server or content router, and the path of routers through which the content is accessed.
Routing advertisements from content servers may also include a measure of the load at that server, specified in terms of the expected response latency. This extra attribute indicates that content which takes longer to access appears ``further away'' from a routing perspective, and may be treated internally by a content router as extra hops in the routing path. The distance this load information is propagated is limited to keep the number of routing updates manageable.
NBRP updates can be authenticated by cryptographic signatures, in a manner similar to Secure BGP [10]. A content server's authenticity is verified by the signature on its initial routing update; content routers receive explicit permission from this content server to advertise routes with their name added to the path list.
Content routers should apply information learned from IP routing to the content routing; if a content peer becomes unreachable then all the content available through that peer is unreachable as well. IP routing information can also be used to select among routes that appear identical at the content routing level. Finally and most importantly, IP routing policies must be consistent with content routing policies so that the decisions made at the content level are faithfully carried out by the IP forwarding level. (It is possible that existing traffic engineering schemes can be used to ensure this behavior; however, we provide some additional ideas on how the two layers can be integrated in the future work section.) Content routers may also make routing decisions based upon information obtained via measurement and mapping techniques.
