On an EVPN L3VPN H-VPN over MPLS, SPEs function as route reflectors (RRs), and UPEs and NPEs function as RR clients. UPEs and NPEs receive specific routes from SPEs.
Create VPN instances on UPE and NPEs and bind the VPN instances to AC interfaces or bind the IPv6 VPN instances to AC interfaces on the UPEs and NPEs.
Configure BGP-EVPN peer relationships between UPEs and SPEs and between SPEs and NPEs. For details, see Establishing a BGP EVPN Peer Relationships.
Configure routing protocols for NPEs and UPEs to exchange routes with CEs. This configuration is similar to configuring PEs and CEs to exchange routes on a BGP/MPLS VPN. For more information, see Configuring Route Exchange Between PEs and CEs or Configuring IPv6 Route Exchange Between PEs and CEs.
Configure SPEs as RRs, and specify UPEs and NPEs as RR clients. For details, see Procedure.
The system view is displayed.
The BGP view is displayed.
The BGP-EVPN address family view is displayed.
The SPE is configured as an RR, and UPEs are specified as its clients.
The SPE is configured to use its own IP address as the next hop of routes when advertising these routes.
To enable an SPE to use its own IP address as the next hop of routes when advertising these routes to UPEs and NPEs, run the peer next-hop-local command with different parameters specified on the SPE for each UPE and each NPE.
One-label-per-next-hop label distribution is enabled on the SPE.
On an EVPN L3VPN H-VPN over MPLS, if an SPE needs to send large numbers of EVPN routes but the MPLS labels are inadequate, configure one-label-per-next-hop label distribution on the SPE.
After one-label-per-next-hop label distribution is enabled or disabled, the label allocated by the SPE for a route changes, which results in packet loss.
The configuration is committed.