Hello AbuRafay63,
The label stack will vary depending on where these traffic engineering or TE tunnels will terminate as this will dictate our label switched path or LSP.
PE to PE
If we are building TE tunnels between PE's then both L2/L3 VPN application traffic will have 2 labels if you were to capture this traffic on the P routers.
------ TE Tunnel -----
| |
PE1 ---- P1 ----- P2 ---- PE2
------------ LSP -----------------
Then the label stack will look like:
Label1: TE label (used to reach PE2)
Label2: VPN/PW label (depending on the service)
In this scenario, the 'P' routers will just label switch the top TE label towards PE2 and then PE2 will just receive the underlying VPN/PW label and forward accordingly.
PE to P
However, if we build TE tunnels between PE's to P routers, then we will see a label stack of 3 labels:
-- TE Tunnel --
| |
PE1 ---- P1 ----- P2 ---- PE2
-------- LSP --------
In this example, in order to reach PE2, we have to use the TE tunnel between PE1 and P2. However, since the TE tunnel terminates at P2, so will our label switching. When we send traffic to PE2 it will only have the TE label which tells us how to reach P2. Hence, the actual LSP terminates on P2 and we will perform normal IP forwarding to PE2 (meaning no labels will be used to reach PE2)
This will break our VPN and PW applications as this will mean the LSP path is not complete towards PE2.
To solve this, we have to go one step further and enable "LDP" on the TE tunnel in order to get the LSP path all the way towards PE2:
LDP
-- TE Tunnel --
| |
PE1 ---- P1 ----- P2 -------- PE2
------ TE LSP --- - LDP LSP -
Now we have an end to end LSP between P1 and P2 and our VPN/PW can flow across our network just fine. And this will make up the 3 labels:
Label1: TE Label (to reach P2)
Label2: LDP Label (To reach PE3)
Label3: VPN/PW label (depending on the service)
For further details please reference this great guide on TE tunnels:
Understanding MPLS TE operations
Thanks,
Victor Hugo Acevedo
