The VPLS: Routed Pseudowire IRB for IPv4 Unicast feature allows a Virtual Private LAN Services (VPLS) multipoint provider edge (PE) device interface to route the Layer 3 traffic along with switch the Layer 2 frames for pseudowire (PW) connections between PE devices. Note that the ability to route frames between interfaces does not affect the termination of a PW into the Layer 3 network (VPN or global) on the same device, or to tunnel Layer 3 frames over a Layer 2 tunnel (VPLS).

In centralized Integrated Routing and Bridging (IRB), only one interface on a PE device is configured with IRB in the domain. All the host devices that are connected to PE devices are configured with this IRB interface IP address as the gateway.

The following figure shows a domain configured with centralized IRB. The figure shows that IRB is configured on the PE device (PE1) interface. All the hosts that are connected to the customer edge (CE) devices (CE1, CE2, and CE3), are configured with the IRB interface IP address (10.10.10.1) as the gateway. In this scenario, only those packets that are destined for the Layer 3 router (30.30.30.0/24) undergo Layer 3 packet rewrite because these interfaces or routers are reachable from the PE1 device. All the hosts communicate only in Layer 2 because they are part of the same bridge domain (10.10.10.x).

In distributed IRB, all the interfaces across all the PE devices are configured with IRB in the domain. The routing protocols enabled on the PE devices allow routes to be learnt between PE devices.

The following figure shows a domain that is configured with distributed IRB. Enhanced Interior Gateway Routing Protocol (EIGRP) is configured on the interfaces of the PE devices (PE1 and PE2), which allows routers (30.30.30.0/24 and 40.40.40.0/24) to exchange routes. Hosts connected to the CE devices are configured with the local IRB interface IP address as the gateway. For example, host 10.10.10.3 is configured with IRB interface IP address 10.10.10.1 as the gateway, and host 10.10.10.4 is configured with IRB interface IP address 10.10.10.2 as the gateway. In this scenario, if the incoming traffic is through a switch virtual interface (SVI), the outgoing traffic can also be reached by SVI through the MPLS network because the relationship is formed across IRB interfaces under the same bridge domain (10.10.10.x).

In the above diagram, where traffic is incoming on PE1 destined for a router interface reachable through PE2, routing takes place on egress of the PE (that is, PE2) based on the gateway configuration. In such a scenario, the packet reaching PE2 always has the source MAC as host MAC, and not the gateway MAC (which ages out after aging time). If the gateway MAC ages out, flooding occurs in the reverse direction traffic. Therefore, we recommend that in case of asymmetric routing, you configure an ARP timeout on the IRB interface that is lower than the MAC aging time so that flooding does not occur across PEs in the VPLS domain.

In this scenario (where traffic is incoming from CE1), both ingress and egress interfaces point to the SVI in the forwarding pipeline of PE1. Although this is expected, it generates ICMP redirect messages. Therefore, we recommend that you configure no ip redirects command on the SVI in interface configuration mode so that ICMP redirect messages are not generated in case of distributed IRB.

The following are the features that are supported on an interface that is configured with the VPLS: Routed Pseudowire IRB for IPv4 Unicast feature:

• IPv4 unicast routing protocols

• Virtual routing and forwarding (VRF)

• DHCP relay

• Address Resolution Protocol (ARP) timeout

• Blocking of Internet Control Message Protocol (ICMP) redirect messages