At this stage we have explored some of the history and technical issues that led to the development of MPLS; we have seen how MPLS works; we have described the MPLS architecture, protocols, and standards; and we have examined the major application areas in which MPLS is currently being used. These applications include traffic engineering, fast rerouting around failures, support for quality of service, scalable virtual private networking, and simplified integration of ATM into IP networks. In these last few paragraphs, we will review the most important characteristics of MPLS and make some observations about the future of the technology.

The most significant technical characteristic of MPLS is that it decouples the information used to forward a packet from the information carried in the IP header. It does this by using labels, rather than anything contained in an IP packet, to make forwarding decisions. This simple fact has far-reaching consequences.

The use of labels to forward packets provides enormous flexibility in the deployment of new routing capabilities. It enables networks to employ routing paradigms other than the traditional hop by hop, destination-based paradigm of IP routing. This leads to the ability to perform traffic engineering, in which the flow of traffic can be flexibly controlled to move traffic from congested paths to paths with excess capacity. It also enables traffic to be quickly rerouted around failed links or nodes without waiting for reconvergence of routing tables among all routers in a network.

By attaching labels to packets, it becomes possible...