Objectives
The simulator ticks each one as soon as the network meets it.
On PE1 create the VPLS instance with ID 100, with peers 2.2.2.2 and 3.3.3.3, and attach GigabitEthernet0/0/1.
On PE2 do the same, with peers 1.1.1.1 and 3.3.3.3.
On PE3 do the same, with peers 1.1.1.1 and 2.2.2.2.
All three PEs have two pseudowires up (display vsi name CUST verbose).
PC1 reaches PC2 and PC3 (ping both from PC1).
Before you start
A VSI holds the peers under pwsignal ldp, and l2 binding vsi joins the customer port. Each peer also needs mpls ldp remote-peer. VPLS needs a full mesh: a frame that arrives on a pseudowire never leaves on another, which is how it avoids loops without spanning tree. A missing peer therefore cuts one pair of sites off. The MAC table shows which site each host is behind.
Concept
VPLS: a virtual switch across an MPLS core
VPLS joins many sites into one Layer 2 segment. Each PE runs a virtual switch (a VFI on Cisco, a VSI on Huawei) with a pseudowire to every other PE, and learns MAC addresses behind its ports and pseudowires. Unknown frames are flooded out of all of them.
Concept
Pseudowires: Ethernet through an MPLS core
A pseudowire makes an MPLS core look like one long cable between two customer ports, called attachment circuits. The ingress PE pushes two labels on each frame: an inner VC label that names the far attachment circuit, and an outer transport label that carries it across the core. LDP signals the VC label between the two PEs, directly, through a session that may cross many routers.
