Configuring BGP RPD

RPD provides a new method of distributing route-policies and allows the NCE to efficiently and dynamically deploy route-policies.

Usage Scenario

In a MAN ingress or IGW scenario, uneven link resource usage or link faults may cause link congestion. To make full use of network bandwidth, you can deploy an inbound traffic optimization solution to adjust route priorities so that traffic is diverted to idle links. In such a scenario, the router functions as a forwarder, and RPD needs to be deployed on it.

In Figure 1, an inbound traffic optimization solution is deployed so that traffic from AS 200 to the backbone network is monitored and scheduled in real time. If the link from Device C to Device A is congested, the traffic enters AS 100 through Device B and reaches the backbone network by path of the PE.

Figure 1 Typical networking of inbound traffic optimization

In this scenario, forwarders receive RPD policies delivered by the NCE and adjust route attributes and advertise routes based on the policies. The traffic optimization policy is configured on the NCE based on the traffic application. The following section describes relevant configurations on the forwarders. For details about how to configure the NCE, see the NCE manual.

Pre-configuration Tasks

Before configuring BGP RPD, complete the following tasks:

Configure basic BGP functions on a device, and establish a connection between the device and NCE to ensure routing reachability between them.

Procedure

Configure basic RPD functions.
1. Run system-view
  The system view is displayed.
2. Run bgp { as-number-plain | as-number-dot }
  The BGP view is displayed.
3. Run rpd-family
  BGP RPD is enabled, and the BGP RPD address family view is displayed.
4. Run peer ipv4–address enable
  BGP is enabled to exchange routing information with the specified peer.
5. Run quit
  Return to the BGP view.
6. Run ipv4-family unicast
  The IPv4 unicast address family view is displayed.
7. Run peer ipv4-address rpd-policy export enable
  The RPD export route-policy function is enabled in the IPv4 unicast address family view.
8. Run commit
  The configuration is committed.
(Optional) Configure GR to prevent the traffic interruption caused by a protocol restart.
1. Run system-view
  The system view is displayed.
2. Run bgp { as-number-plain | as-number-dot }
  The BGP view is displayed.
3. Run graceful-restart
  GR is enabled in the BGP view.
4. Run rpd-family
  The BGP RPD address family view is displayed.
5. Run peer ipv4-address graceful-restart static-timer restart-time
  GR is enabled in the BGP RPD address family, and a GR restart timer is set.
6. Run commit
  The configuration is committed.
(Optional) Configure router ID-based filtering on non-RRs in the RR scenario so that the non-RRs only accept the RPD routes that match the router ID of the local BGP process. If a large number of RPD routes are accepted, a large number of policy nodes are generated accordingly, which reduces the performance.
1. Run system-view
  The system view is displayed.
2. Run bgp { as-number-plain | as-number-dot }
  The BGP view is displayed.
3. Run rpd-family
  The BGP RPD address family view is displayed.
4. Run router-id filter
  Router ID-based filtering is enabled.
(Optional) Configure a delay for protocols to apply an updated RPD route-policy if the original policy changes.
1. Run ipv4-family unicast
  The IPv4 unicast address family view is displayed.
2. Run rpd-policy change notify-delay delay-time
  A delay is configured for protocols to apply an updated RPD route-policy if the original policy changes.

Verifying the Configuration

After configuring BGP RPD, verify the configuration.

Run the display bgp peer ipv4-address rpd export-policy command to check RPD route-policies.
Run the display bgp rpd routing-table command to check information about RPD routes.