I am labbing out a scenario I want to implement and not sure if I'm being stupid or if the simulation tool is just not playing along. I am using Huawei's eNSP which is notorious for its bugs - not that the provider matters, I'm here to see if my conceptual understanding is correct.
Please view the diagram attached as I may not be the best at explaining the topology.
My scenario involves having two border routers (BR01 and BR02), BR01 connecting to an internet exchange (IX) point, these border routers will peer to each other using their loopback IPs which are advertised by an IS-IS/MPLS LDP backbone.
The backbone MPLS routers should only have core IPs, i.e. the loopbacks in their routing tables - unlike the border routers which will have full internet routing tables. These loopbacks are advertised via IS-IS within the MPLS backbone and LSPs are created through the use of MPLS LDP.
My understanding is that if an LSP exists from BR01 to the loopback of BR02 and vice versa the traffic should simply be label-switched until it reaches the loopback.
I do have next-hop self(local in Huawei) on these BGP peers allowing advertised routes to have their next hops set as the loopback address of the destination border.
The issue I am experiencing takes place when an IX01 external network, tries to reach a network initially advertised by BR02, that BR01 learns from BR02 and then eBGP re-advertises to IX01.
The traceroute takes me from IX01 to BR01 and then dies there. When I run a packet capture on the interface between BR01 and Backbone 01 I see the original source and destination IPs with no MPLS header for label switching.
What I currently think is that the MPLS label header should be in that packet as BR01 sees the next-hop as the loopback of BR02 of which it does have a valid LSP for.
If looking at the diagram, I am sourcing from a PC with IP 41.0.0.2/24 GW 41.0.0.1 which resides behind the IX router. This network is eBGP advertised to BR01 which then iBGP advertises it to BR02, where the destination network 42.0.0.0/24 is advertised back with next-hop self. Destination PC being 42.0.0.2
Am I being stupid? All the reading I'm doing makes this task look trivial but I'm just not getting it right in my lab
No comments:
Post a Comment