Question about Asymetric routing on VPLS and MPLS seperate edge routers
Hi guys, for example.
I am making this home lab on gns3. It is almost perfect but when I enable both WAN cloud routers VPLS and MPLS (just a router with BGP AS on it only to mimic cloud)
I have 2 routers (vpls edge) and (MPLS edge) at each LAN site coming from a main CORE router. I have the IGP as EIGRP and the cloud EGP as EBGP.
so if I shut down the vpls port on the core router, the traffic can trace there but cannot get back because the vpls edge router path selection at each lan has a metric of BGP redistributed into EIGRP one day and it tries to go out the core and then the vpls router and not back the same path it came on?. The VPLS router down at my original site I am playing with a failover scenario and the bgp does not replicate that the vpls route back to me is dead. Long time ago they had a transit cable in between routers but that is now gone. So how can I make the packet return the same path it came from (why? because the path back on the vpls (preferred route always) is dead? this is asymetric routing and poor route logic? how can i overcome this? and make the packet return the same path it came?
any ideas guys ? thank you.
I am making this home lab on gns3. It is almost perfect but when I enable both WAN cloud routers VPLS and MPLS (just a router with BGP AS on it only to mimic cloud)
I have 2 routers (vpls edge) and (MPLS edge) at each LAN site coming from a main CORE router. I have the IGP as EIGRP and the cloud EGP as EBGP.
so if I shut down the vpls port on the core router, the traffic can trace there but cannot get back because the vpls edge router path selection at each lan has a metric of BGP redistributed into EIGRP one day and it tries to go out the core and then the vpls router and not back the same path it came on?. The VPLS router down at my original site I am playing with a failover scenario and the bgp does not replicate that the vpls route back to me is dead. Long time ago they had a transit cable in between routers but that is now gone. So how can I make the packet return the same path it came from (why? because the path back on the vpls (preferred route always) is dead? this is asymetric routing and poor route logic? how can i overcome this? and make the packet return the same path it came?
any ideas guys ? thank you.
Comments
-
daveyb
Member Posts: 28 ■□□□□□□□□□
You're going to need to provide a diagram.
Asymmetric routing in general is not a problem. You will be hard pressed to find a path between two nodes in the internet that do not have an asymmetrical path between them.
How is your VPLS router configured? It seems odd that it never gets detected as down. -
itdaddy
Member Posts: 2,089 ■■■■□□□□□□
each site has vpls and mpls router. bgp is redistributed from the WAN clouds into EIGRP. vpls bgp metric is best path metric redistributed into EIGRP while the mpls is the worst path metric but when vpls is down say on the originating site, it still choose the vpls route on the opposite end back when it should take mpls back
since the vpls router is down. i know this works, i have seen it but just trying to lap it up. granted I could
use ip sla ping and then have it inject a route in L3 table or do floating route provided only if the Dallas Tx vpls reports a bad route fast enough? what are my options that you have seen in real world. Thank you.
I agree aymetric routing happens all the time, but in this case kind of different but real world nonetheless. thank you bud
Diagram below in clink.