This feature implements RFC3107 that allows carrying a label stack with BGP route updates, using multi protocol BGP.

To understand why BGP-LU tunnels are required to be redistributed into LDP and vice versa, let’s consider the following basic topology for InterAS Option C. It uses IS-IS as IGP where indicated along with LDP:

Normally the ingress router in the following diagram has no control over an autonomous system border router’s

BGP-LU 4.25.0F EPE

Egress Peer Engineering (EPE) using BGP LU enables traffic engineering of the links between Autonomous Systems

BGP-LU EPE 4.27.0F

The MPLS ping/traceroute utilities were extended in EOS version 4.26.0F to support connectivity checking of BGP

Normally BGP allocates local labels and installs LFIB entries for all received IPv4/IPv6 Labeled Unicast (LU) routes in anticipation of readvertising them with nexthop-self. However, some deployments don’t require nexthop-self with LU routes, so LFIB hardware resources are needlessly allocated, which can present an issue in large scale LU deployments. 

BGP-LU LFIB EOS 4.30.2F