IP Routing

How BGP Works

BGP’s underlying algorithm is the simple DV protocol -- when a BGP speaker hears a prefix via multiple paths, it selects the “best” path for insertion in the routing table and announces this “best” path to other peers.

We are already familiar with the DV protocol via RIP and IGRP. However, unlike RIP and IGRP, BGP’s purpose is inter-AS routing, which is a different beast from intra-AS routing. The architects of BGP created several new structures to support inter-AS routing. This section gives an overview of these new structures.

Let’s start at the beginning. A unique, registered AS number is required for the BGP process to connect to the Internet. Then, unlike IGPs, BGP does not contain any mechanism for automatic neighbor discovery. The network administrator must manually define BGP neighbors. This is appropriate given that the neighbor may be in another AS.

Protocols such as RIP and OSPF are generous in exchanging updates; the network statement permits all known subnets to be announced in updates. BGP operates under a different paradigm -- updates should be tightly controlled. Cisco’s implementation of BGP gives several methods to control not only the prefixes that are announced but also the associated attributes.

As the Internet routing protocol, BGP must support a very large routing table: the current size of the Internet routing table is roughly 70,000 prefixes. Given this size, periodic refreshing (such as every 30 s in a RIP network) of this table (with the associated ...