How Does MPLS Routing Work? A Complete Guide and FAQ for MPLS Routing

What does MPLS stand for? MPLS stands for Multi-Protocol Label Switching.
What is MPLS? It is a mechanism for routing traffic within a telecommunications network, as data travels from one network node to the next. MPLS can provide applications including VPNs (Virtual Private Networks), traffic engineering (TE) and Quality of Service (QoS).

How does MPLS work? In MPLS, packets are directed through the network based on  an assigned label. The label is associated with a predetermined path through the network, which allows a higher level of control than in packet-switched networks. MPLS routing allow differing Qos characteristics and priorities to be assigned to particular data flows, and operators can predetermine fallback paths in the event that traffic must be rerouted.
With pure IP (Internet Protocol) routing in a packet-switched network, each data packet could determine its own path through the network – which was a dynamic flow, but not predictable. However, it was very cost effective.
In previous circuit-switched telecom networks, physical wires and T1 lines carried data and voice traffic. That provided predictable routes, but was very expensive and difficult to scale because of the need to put in extensive infrastructure.
So MPLS and a similar technology, Carrier Ethernet (CE) have evolved to allow control of network routing, creating paths that act like a point-to-point connection within the network, but are virtual and flexible instead of physical.
MPLS has the “touch and feel attributes of being able to set up – not physical circuits, but virtual circuits. Ethernet virtual circuits,” said Errol Binda, senior marketing director of Aviat Networks.
What is the “label switching” in MPLS routing?
As packets travel through the MPLS network, their labels are switched or swapped.
The packet enters the edge of the MPLS backbone, is examined, classified and given an appropriate label, and forwarded to the next hop in the pre-set Label Switched Path (LSP). As the packet travels that path, each router on the path uses the label – not other information, such as the IP header – to make the forwarding decision that keeps the packet moving along the LSP.
However, within each router, the incoming label is examined and its next hop is matched with a new label. The old label is replaced with the new label for the packet’s next destination, and then the freshly labeled packet is sent to the next router. Each router repeats the process until the packet reaches an egress router.
The label information is removed at either the last hop or the exit router, so that the packet goes back to being identified by an IP header instead of an MPLS label.
What are the differences between Carrier Ethernet and MPLS?
“They both have a labeling mechanism,” said Binda. “They both have the ability, because you’re creating headers, to create virtual tunnels, Ethernet virtual circuits. … The primary difference between the two comes in terms of how rich of networking you have, and how specific the configuration of the label or paths are.”
In general terms, MPLS has more sophisticated ability to control network dynamics and greater scalability than carrier Ethernet. Both have advantages in cost reduction per bit of data, as compared to traditional transport mechanisms.
“They actually can complement one another,” Binda added. Some carriers design their networks so that carrier Ethernet is in a portion of the network, and MPLS is in another portion. MPLS can also be designed so that it reaches to the network edge.
“Everybody wants to move away from legacy networks toward carrier Ethernet and MPLS,” Binda said. “The industry as a whole is thinking about how to make these technologies work together.”