BGP Route Distinguisher (RD) vs Route Target (RT): What They Do and How to Troubleshoot

Short answer: A BGP Route Distinguisher (RD) makes otherwise identical VPN routes unique; a Route Target (RT) is the extended-community policy tag that controls which VRFs import or export those routes. RD answers “which route is this?”; RT answers “where should it go?” [1][2]

Why RD and RT are often confused

In an MPLS L3VPN, separate customers can use overlapping IPv4 prefixes. A plain prefix such as 10.20.0.0/16 is ambiguous in a shared BGP VPN address family. The PE prepends an RD to form a VPN-IPv4 NLRI, allowing BGP to carry distinct routes for the same IPv4 prefix. The RD does not itself identify a customer or decide distribution. [1]

An RT is carried as an extended community. Export policy attaches one or more RTs to a route; a VRF's import-RT policy makes matching routes eligible for that VRF. This is the control-plane membership mechanism used to build hub-and-spoke, shared-services, and other VPN topologies. [1]

How route processing works

  1. A PE learns a customer prefix in a VRF.
  2. It represents the route in VPN BGP with an RD, making its NLRI distinct, and attaches configured export RTs.
  3. Remote PEs receive the VPN route and compare its RTs with VRF import policy.
  4. A matching route may be installed in the VRF, subject to other routing and next-hop decisions.

Think of the RD as a route-identity namespace, and the RT as a distribution label. Changing an RD alone does not create connectivity between VRFs; changing RT import/export policy can change route membership. [1]

Configuration example (Cisco IOS-style)

vrf definition CUSTOMER-A
 rd 65000:101
 address-family ipv4
  route-target export 65000:101
  route-target import 65000:101
 exit-address-family
!
vrf definition SHARED-SERVICES
 rd 65000:900
 address-family ipv4
  route-target export 65000:900
  route-target import 65000:101
 exit-address-family

Here, SHARED-SERVICES imports CUSTOMER-A-tagged routes, while CUSTOMER-A does not automatically import the shared-services routes. For two-way communication, design the reverse import/export policy deliberately. Exact syntax and defaults vary by platform and release; validate against the vendor's documentation before deployment.

Traffic engineering: what RD/RT do and do not control

  • Outbound: RT import policy determines which VPN routes are eligible to appear in a local VRF, shaping reachable destinations. It does not choose the best path by itself; normal BGP/IGP selection still applies.
  • Inbound: An RT can influence which remote VRFs learn a route, indirectly influencing where traffic can be routed. It does not force a remote network to send traffic to a particular PE or ISP.
  • Cannot control: RD is not a path-preference knob, a security boundary by itself, or a forwarding label. RT is not packet encryption, firewall policy, or guaranteed isolation if the surrounding import/export policy is wrong. Data-plane reachability and security controls remain necessary.

EVPN/VXLAN and data-center use

EVPN also uses RDs to distinguish routes advertised by a MAC-VRF and RTs to control route distribution among EVIs. RFC 7432 specifies an RD per MAC-VRF and one or more RTs; RTs may be configured or derived in certain deployment models. The same conceptual separation applies: RD uniqueness versus RT membership. [2]

In a BGP EVPN/VXLAN fabric, RT import/export policy commonly maps tenant or bridge-domain routing information between VTEPs. Cisco ACI and other policy-based fabrics may expose different abstractions and automation workflows, so do not assume a classic VRF configuration maps one-to-one onto their GUI or fabric policy model.

Verification

show bgp vpnv4 unicast all
show bgp vpnv4 unicast vrf CUSTOMER-A
show vrf detail CUSTOMER-A
show ip route vrf CUSTOMER-A

Confirm the route NLRI includes the expected RD, the path carries the intended RT extended community, and the destination VRF imports that RT. Then confirm the prefix appears in the VRF BGP table and routing table, and test forwarding to the next hop. On FRRouting, inspect the BGP VRF/address-family configuration and advertised/received routes using the available show bgp commands for that release. [3]

Troubleshooting checklist

  • Is the route present at the source PE and advertised in the VPN address family?
  • Is the RD unique where the platform requires it, especially across MAC-VRFs on a PE in EVPN?
  • Does the advertised route carry the expected export RT?
  • Does the destination VRF import that exact RT (including ASN/administrator and value)?
  • Are route-reflector address-family, extended-community propagation, and route-target constraints configured correctly?
  • Does the route have a resolvable next hop and pass route-policy filters?
  • Does the VRF route table contain the route, and do data-plane labels/tunnels and firewall policies permit traffic?

Interview answer

“An RD makes overlapping VPN prefixes unique in the BGP VPN address family. An RT is an extended-community policy tag that controls which VRFs are eligible to import a route. RD identifies; RT distributes.”

Related reading

Sources

[1] RFC 4364 — BGP/MPLS IP VPNs
[2] RFC 7432 — BGP MPLS-Based Ethernet VPN
[3] FRRouting BGP documentation
[4] Cisco BGP-VPN Distinguisher documentation

Try it: BGP Route Lookup - free, runs in your browser.

Comments

Popular posts from this blog

BGP Neighbor States

Reason legend in show ip ospf statistics