VPLS resiliency and redundancy
Domain 5 tests VPLS resiliency and redundancy: how Nokia designs prevent loops, provide dual-homed access, and recover from pseudowire or transport failures.
Mechanisms
- VPLS multi-homing uses BGP-AD-based election to choose a designated forwarder for a multi-homed site. The standby PE does not forward for that site, preventing a loop through the VPLS.
- MC-LAG dual-homes a CE to two PEs with a LAG. The CE sees one logical link; the PEs coordinate forwarding. This is different from VPLS multi-homing.
- Pseudowire redundancy can use a primary and standby spoke SDP. The standby becomes active on failure; check revertive behavior.
- Transport redundancy uses redundant LDP/RSVP paths, IGP convergence, and BFD to detect failures quickly.
- Split horizon remains mandatory in a mesh to prevent pseudowire loops.
Rules
- Do not attach the same CE site actively to two PEs without a loop-prevention mechanism. Two active SAPs into one VPLS can create a Layer 2 loop.
- Verify the designated forwarder, standby SAP state, and pseudowire active/standby state after failover.
- Redundancy is only useful if the standby path is operational; monitor SDP, LDP, and BFD state.
Worked example
CE-A is dual-homed to PE1 and PE2, and both PEs have SAPs in VPLS 100. If both SAPs forward, a broadcast from CE-A can loop back through the VPLS. With VPLS multi-homing, one PE is elected designated forwarder and the other keeps its SAP in standby for that site. With MC-LAG, the CE's LAG is active on both links but the PEs coordinate so the CE does not see a loop. The choice depends on whether the CE supports LAG.
Traps
Treating MC-LAG and VPLS multi-homing as the same; forgetting that split horizon does not solve a dual-homed CE loop; assuming a standby pseudowire is healthy without testing failover; and ignoring revertive behavior.