VPLS redundancy access design

[suneq]

Hi all,

I want to connect our CE (a stack of 2 x C9500) to 2 PE VPLS. This is my first time with VPLS and I have some doubts to clarify.

I suppose that I cannot configure 2 uplink ports as L3 port without specific configuration on the PEs (cause I simply cannot configure the same subnet on 2 different ports, some kind of MLAG will be required -- am I correct?) Therefore, I plan to configure them as L2 ports with a SVI. I suppose that in this design, STP will block 1 of the 2 uplink; the STP convergence time might not be great but this active-passive design simply works without any specific problem/configuration -- again, am I correct?

Thanks for your advices.

[Leonid Voronkin]

Hi. Am I understood correctly. You are customer. In your area of responsibility only CE and you won't configuring vpls. If so you don't need to think about anything. Just make order on Multi-homing VPLS with LAG and configure LAG from your side. Actually from your point of view there is no MC-LAG or any others technologies for redundant links.

If you want to understand how does it works read this document section 4.4.7 (especially 4.4.7.4)

https://www.cisco.com/c/en/us/support/docs/routers/asr-9000-series-aggregation-services-routers/116453-technote-ios-xr-l2vpn-00.html#anc43

[suneq]

Hi Leonid,

Thanks for your reply.

It will be easier with a diagram : the as-is architecture is on the left and the to-be architecture on the right.

As you can see on the left, my client has 2 SW connected to 2 different PE VPLS. The PE-facing interfaces are configured as L3 ports and they are working properly. SW1, SW2 and the remote site can ping each other (10.0.0.0/24 subnet)

Now for several reasons, the client wants to stack the 2 switches. We checked with the provider and they do not support LAG so we cannot configure Port-Channel on our side. 

As shown on the right of the diagram, I think we can simply configure Gi1/0/1 and Gi2/0/1 as L2 access port (they were L3) and then configure a SVI.

As the pseudowire between PE1 and PE2 can transport BPDU, I suppose that the port 2/0/1 of the Stack will be blocked by STP but apart from that, it should work fine:

1. from SVI 10.0.0.1/24 I should be able to ping the remote site 10.0.0.3/24
2. if the link between PE1 and Stack fails, port 2/0/1 will no longer receive BPDU and will be FWD, the reconvergence with RSTP should be quick enough.

I read quickly the article you shared and my assumption seems correct. Any confirmation/comments will be highly appreciated.