Page 359

arge BGP networks have several methods for managing large

BGP networks, such as using filters, configuring private AS numbers (ASNs),
creating peer groups, creating confederations, and using route reflectors. All
of these methods are quite complex. In this chapter, we will learn more about
all of these problems.

When a router reaches more than 100 BGP sessions running concurrently,

most network administrators fluent in BGP recommend that you configure
route reflectors. When route reflectors are used, a router needs to become a
peer only with a route reflector instead of with each individual router. The
route reflector's responsibility is to maintain a routing table for all internal
peers connected to the reflector. The route reflector can collect the same
number of routes that a router can learn from a full mesh.

You can use confederations to control the size of interior BGP meshes and

allow an AS to be broken up. Several sub-AS numbers can be configured
inside of a real AS. Sub-ASes must use the private reserved ASNs. All of the
routers within a sub-AS must be fully meshed, and one of the routers in a
sub-AS creates a peer with one of the routers from the other sub-AS. This
enables the overhead on each router to be reduced and a complete BGP table
to be shared to the entire AS.

Before we look at the more advanced features just described, let's first

take a look at different methods of applying policies to the routes advertised
by BGP, such as using distribute lists, route maps, and prefix lists.

www.sybex.com