How To Implement Bgp; Enabling Bgp Routing - Cisco NCS 6000 Series Configuration Manual

How to Implement BGP

A route-policy is mandatory to filter the imported routes. This reduces the risk of unintended import of routes
between the Internet table and the VRF tables and the corresponding security issues.
There is no hard limit on the number of prefixes that can be imported. The import creates a new prefix in the
destination VRF, which increases the total number of prefixes and paths. However, each VRF importing
global routes adds workload equivalent to a neighbor receiving the global table. This is true even if the user
filters out all but a few prefixes. Hence, importing five to ten VRFs is ideal.
How to Implement BGP

Enabling BGP Routing

Perform this task to enable BGP routing and establish a BGP routing process. Configuring BGP neighbors is
included as part of enabling BGP routing.
Note At least one neighbor and at least one address family must be configured to enable BGP routing. At least one
neighbor with both a remote AS and an address family must be configured globally using the address family
and remote as commands.
Before you begin
BGP must be able to obtain a router identifier (for example, a configured loopback address). At least, one
address family must be configured in the BGP router configuration and the same address family must also be
configured under the neighbor.
Note
If the neighbor is configured as an external BGP (eBGP) peer, you must configure an inbound and outbound
route policy on the neighbor using the route-policy command.
Note
While establishing eBGP neighborship between two peers, BGP checks if the two peers are directly connected.
If the peers are not directly connected, BGP does not try to establish a relationship by default. If two BGP
peers are not directly connected and peering is required between the loop backs of the routers, you can use
the ignore-connected-check command. This command overrides the default check that BGP performs which
is to verify if source IP in BGP control packets is in same network as that of destination. In this scenario, a
TTL value of 1 is sufficient if ignore-connected-check is used.
Configuring egp-multihop ttl is needed when the peers are not directly connected and there are more routers
in between. If the egp-multihop ttl command is not configured, eBGP sets the TTL of packets carrying BGP
messages to 1 by default. When eBGP needs to be setup between routers which are more than one hop away,
you need to configure a TTL value which is at least equal to the number of hops between them. For example,
if there are 2 hops (R2, R3) between two BGP peering routers R1 and R4, you need to set a TTL value of 3.
Routing Configuration Guide for Cisco NCS 6000 Series Routers, IOS XR Release 6.4.x
46
Implementing BGP