I am trying to make the H-VPLS interops with Juniper router in a Hub and Spoke config where Spoke connects with two HuB. Full mesh VPLS is just working fine.
We did some extensive tests and found that the HuB (Juniper) router DO NOT receive the label for the particular VPLS circuit from the Spoke (Extreme) router when "l2vpn vpls” PW is used. Which means the Spoke DO NOT send the label at all.
#configure l2vpn vpls TEST add peer xxx.xxx.xxx.100 core primary
But, the Spoke (Extreme) router DOES sends the label when the "l2vpn vpws” PW is used and connection gets established. However the Spoke receives the labels from the HuB site all the time with both PW type.
#configure l2vpn vpws TEST add peer xxx.xxx.xxx.xxx core
While Extreme as HuB and Spoke we specifically attach the vpls instance with the peer and define the H-VPLS MTUs the peer as Spoke in a HuB .
#configure l2vpn vpls TEST add peer xxx.xxx.xxx.xxx spoke
I suspect that the parameters “spoke” enables the HuB to send some probes signaling bits to the Spoke peer to notify its availability as a HuB. Once Spoke receives the bit probe from the HuB then only it advertise its labels to the HuB. Perhaps this methodology is used for multihoming capability for the Spoke router i.e. one Spoke connecting with two different HuB sites for redundancy with "l2vpn vpls” PW on primary and secondary HuB.
Is this some kind of propriety implementation? Does Extreme follows any RFC standards for this implementation? Anyone tried this before?
