Information Technology Reference
In-Depth Information
Convergence
is the process in which a network synchronizes to the same routing informa-
tion after a change in the network. A network that is not converged can lead to packet loss
or forwarding loops. However, stability can be reduced if a network is in a constant state of
convergence. A proper balance of stability and convergence can be dependent on the ser-
vices a network provides. For example, when BGP is used to provide virtual private net-
work (VPN) services over a shared Multiprotocol Label Switching (MPLS) network, there
might be more emphasis on convergence. Chapter 10, “Multiprotocol BGP and MPLS
VPN,” provides detailed discussions of this subject. The discussion of BGP convergence
tuning is presented in detail in Chapter 3.
Scalability
You can evaluate BGP's scalability in two areas: the number of peer sessions and the
number of routes. Depending on the configuration, hardware platform (CPU and memory),
and Cisco IOS release, BGP has been proven to support hundreds of peer sessions and to
maintain well over 100,000 routes.
Several measures are available to increase BGP scalability. These measures reduce either
the number of routes/paths to be maintained or the number of updates to be generated.
As a form of distance vector protocol, BGP updates its peers only with the paths it uses. In
other words, only the best paths are advertised to its peers. When the best path changes, the
new path is advertised, which lets peers know to replace the previous best path with the new
best path. This action is an implicit withdrawal of the previous best path.
When BGP is used to exchange reachability information within the same AS, all BGP speak-
ers are required to be fully meshed. Because fully meshed networks tend to limit scalability
because of the number of sessions that must be maintained on each router and the number
of updates that must be generated, route reflection and confederation are two methods that
increase the scalability of BGP networks. Detailed discussions of these methods are included
in Chapter 7, “Scalable iBGP Design and Implementation Guidelines,” Chapter 8, “Route
Reflection and Confederation Migration Strategies,” and Chapter 10.
Aggregation of routes is another tool that BGP uses to reduce the number of prefixes to be
advertised and increase stability. Proper aggregation is, in fact, a requirement in the Internet,
as discussed in Chapter 6, “Internet Connectivity for Enterprise Networks.”
Lowering the number of updates to be generated reduces CPU utilization and enables faster
convergence. In IOS, peers that have the same outbound policy can be grouped in a peer
group or update group. One update is generated and then replicated for the entire group.
The subject of performance improvement using update grouping is discussed in detail in
Chapter 3.
