MPLS: Two Customer Sites Over a Provider Core

The scenario

Site-A (192.168.1.0/24) and Site-B (192.168.2.0/24) belong to one customer. Each site has a customer edge router, CE1 and CE2, which hands traffic for the other site to the provider's edge router, PE1 or PE2. Between the PEs is the provider's core router, P. PE1 pushes label 17001 onto traffic for 192.168.2.0/24, P swaps it for 17002, and PE2 pops it and forwards the plain packet to CE2. The other direction uses 17010, swapped to 17011 at P and popped at PE1. Every packet carries exactly one label. P has routes only to its own two links, so the customer's 192.168 addresses exist only on the CE and PE routers. No router runs OSPF; all routes and labels are static.

What you will work out

See how a provider carries a customer's traffic across a core that knows nothing about the customer's addresses, and what a full MPLS L3VPN adds on top.

Why this matters

MPLS L3VPNs are a common way for enterprises to connect branch offices over a provider's network. The property this lab shows, a core that forwards customer traffic by label without knowing the customer's addresses, is what lets one provider core serve many customers. When you meet the full service, the extra pieces are VRFs to keep customers apart on each PE, a VPN label to pick the right VRF at the far end, and MP-BGP to share the routes.

Running it

MPLS: Two Customer Sites Over a Provider Core opens on a canvas, already wired and addressed. Trace a packet through it hop by hop, or run the same topology live on real Linux network namespaces. How the two modes differ.