Chapter 15. The Future of MPLS

As service providers proceed along the path to deploying their Next Generation Network (NGN) networks and services, IP/MPLS is and will be a key architectural component of voice, video, and mobile data services for these NGN networks. This chapter discusses the future of MPLS and the role of IP NGN and MPLS. Finally, two experts from the industry who have been active with MPLS technology from the beginning offer a view on the future of MPLS. First is George Swallow, distinguished engineer at Cisco Systems and Internet Engineering Task Force (IETF) MPLS Working Group co-chair. Second is Adrian Farrell, who has been active at the IETF, particularly in Common Control and Measurement Plane (CCAMP) where Generalized MPLS (GMPLS) is discussed. Is there life after MPLS? The conclusion is left to you.

You have seen that MPLS packets can run over a link layer and a link layer packet (Ethernet, ATM, Frame Relay (FR), Point-to-Point Protocol (PPP), and so on) can also be run over MPLS. You have also seen that Layer 3 (L3) packets run on MPLS and MPLS frames can be run over IP packets. This implies that MPLS is neither a Layer 2 protocol nor a Layer 3 protocol. In fact, Yakov Rekhter, one of the inventors of MPLS has said, "MPLS doesn't fit in the OSI layer at all." So the question must be asked, "Where does MPLS fit, and what is its future?"

MPLS is emerging as a widely acceptable technology for network convergence. Evolving from Cisco tag switching, MPLS has become the defacto standard for delivering services on IP networks. More than 250 Cisco customers, service providers, and large enterprise customers have deployed MPLS for Layer 3 VPNs, traffic engineering, or Layer 2 VPNs.

As noted, MPLS follows a simple paradigm of control and forwarding plane separation. We have also seen that the forwarding plane uses simple shim headers as labels. The complexity is entirely in the control plane w.r.t label distribution and management.

This simple paradigm of control and forwarding planes can easily be extended to other transport networks, such as ATM, Frame Relay, and optical networks. In fact, the ATM MPLS networks has not been discussed in this book due to the fact that fewer and fewer deployments exist, but ATM MPLS networks have been around for more than five years. As shown previously in Chapter 3, "Technology Overview," the label is carried in the Virtual Path Identifier/Virtual Channel Identifier (VPI/VCI) field of the ATM cell in the case of ATM MPLS networks. Similarly, the label can be carried in the Frame Relay data-line connection identifier (DLCI) field for a Frame MPLS network. Remember that with ATM MPLS and Frame MPLS, references are not to networks running either MPLS over ATM or MPLS over Frame Relay, but rather to the setup of virtual connections (LSP) using MPLS as a control plane with each Frame Relay switch and ATM switch behaving as a Frame Relay Label Switch Router (LSR) and ATM LSR.

One other important area to which this concept can easily be extended is the time-division multiplexing (TDM) and optical space. For example, by designating channel numbers or optical tributaries as labels and using a GMPLS control plane to set up the cross connects of channels, you can set up a label-switched path (LSP) in a TDM or an optical network. This cross connection of TDM channels or setup of optical paths across the network is analogous to a packet LSP that is set up across the LSR. In this chapter, we explore in detail this generalization of MPLS, known as GMPLS and find out how this can help integrate IP and optical networks and bring efficiency in provisioning Layer 1 circuits.

Also discussed in this chapter are other applications and enhancements to MPLS that help build other services in the future such as guaranteed bandwidth multicast for video distribution, ubiquitous transport of MPLS VPNs across IP networks, and dynamic setup of security adjacency for better encryption in the MPLS VPNs environment.