Border Gateway Protocol - CCIE Routing and Switching Exam Certification Guide - page 447

Route Dampening
Route dampening minimizes the instability caused by oscillating or flapping routes. Route flaps
occur when a valid route is considered invalid and then valid again. Route dampening is a
method to stop unstable routes from being propagated throughout an internetwork. The BGP
dampening process assigns a penalty to a route for each flap. The penalty carries a value of
1000. Penalties are cumulative. The route is suppressed if the penalty reaches the suppress limit,
which is 2000 by default.
The penalty value decreases by half when the configured half-life time is reached (15 minutes
by default). When the value reaches the reuse value, it becomes active. The maximum time a
route can be suppressed is 60 minutes by default. The maximum-suppress-time field in the
command changes the maximum time the route can be suppressed.
The command syntax to enable BGP route dampening is the following:
router bgp as-number
bgp dampening half-life-time reuse-value suppress-limit maximum-suppress-time
The half-life-time is the number of minutes for half-life time for the penalty and is a number
between 1 to 45, inclusive. The default half-life-time is 15 minutes. The reuse-value identifies
the value to which the penalty needs to fall before the route can be re-advertised to the network
and is a number from 1 to 20,000. The default reuse-value is 750. The suppress-limit indicates
the value the penalty needs to reach before the route is suppressed; it is a number from 1 to
20,000. The default suppress-limit is 2000. The maximum-suppress-time is the maximum dura-
tion to suppress a stable route; values are from 1 to 255. The default maximum-suppress-time is
four times the half-life or 60 minutes.
BGP Peer Groups
Peer groups are a grouping of BGP neighbors that are applied to the same policies. The policies
can be the following: distribution list, route maps, next hop, update source, and so on. Peer
groups can be assigned to iBGP peers or eBGP peers, but a peer group cannot contain both
types of peers. The major benefit of a peer group is that it reduces the amount of CPU and mem-
ory resources on the router because it allows the routing table to be looked up once for update
generation for all routers in the peer group. For large tables with a large number of prefixes, this
can significantly reduce the load. Another benefit is that it simplifies the configuration of BGP
neighbors because each command policy can be configured for the peer group without having
to configure the policy for each BGP peer. For example, 15 BGP neighbors with 5 command
policies are configured with approximately 75 commands. If all BGP neighbors are assigned to
a peer group, the number of commands is reduced to approximately 20.
Example 9-14 shows the configuration of the iBGP peer group globalnet. All peers in the group
share the same neighbor globalnet commands. The globalnet peer group is defined with the
neighbor globalnet peer-group command. The shared commands are configured for the peer
group; in this example, the remote AS is 65500, the update-source is loopback 0 of each router,
each router announces itself as the next hop of its advertised routers, the BGP version is 4, and