MPLS and Next-Generation Networks: Foundations for NGN and Enterprise Virtualization

This section focuses on MPLS technology as a service building block and foundation for enterprise virtualization implementation.

MPLS offers the following benefits for service providers and enterprises:

• Flexible classification of packets and the optimization of network resources.

• Label distribution through various protocols such as BGP, LDP, RSVP, and PIM.

• The coexistence of different distribution protocols in the same LSR.

• The redundancy of numbering and global label allocation, as labels that have only a local significance.

• The introduction of modular value-added applications such as traffic engineering, quality of service, multicast, and VPN.

• Facilitation of the evolution of legacy services via Any Transport over Multiprotocol Label Switching (AToM) and even the introduction of Layer 2 VPNs as the cost of retaining Frame Relay and ATM infrastructures becomes prohibitive.

• Unification of optical and routing control planes in GMPLS to evolve SDH and Sonet services. Also, GMPLS is used to generalize the MPLS control plane over many types of transports, including packet-type networks.

MPLS, therefore, provides the predictability of routing performance required to support differentiated services and the capability to offer tight SLAs associated with these differentiated service constructs. MPLS facilitates the integration of multiple services over a common switching platform, therein contributing to the reduction of operating expense. MPLS traffic engineering can reduce the management burden for IP-based services via the creation of backup paths and by facilitating the deployment of VoIP VPNs.

Path diversity can result in unpredictability in end-to-end delay because the number of links and routers by successive packets can be varied. With path diversity, each router must perform a full routing table lookup to determine the next-hop router along the path. This process is time-consuming and produces difficulties in attaining end-to-end delay within acceptable bounds for voice and video applications.

MPLS addresses the problem in several ways. Label-switched networks fundamentally implement a simpler procedure to determine the exit path for any incoming packet (as previously discussed). In addition, traffic can be fixed to certain paths (constrained routing) via traffic engineering, which allows the service provider to exert more control over traffic congestion. For resiliency, the service provider can create backup paths such that in the event of a link or node failure, the alternative path can be activated to reduce service failure. Therefore, MPLS opens up new possibilities for traffic engineering. The definition of LSPs and their FECs allows specific traffic flows to follow paths that deviate from the shortest path designated by classical IP routing protocols.

Implementing the DiffServ architecture with MPLS can provide traffic CoS capabilities over a packet-based network, therein providing the capability to deploy voice and multimedia applications marked with a service priority. Service providers can also deploy MVPNs to support applications using streaming, such as IPTV, Windows Media Player, Real Player, Quick-Time Video Conferencing, and Netmeeting.

Service providers are deploying Layer 2 VPNS to reduce TDM switching and transmission costs as AToM technology emulates Layer 2 services, such as Frame Relay, ATM, PPP, HDLC, and Ethernet. Further, Fast Reroute is used to provide network resilience in place of SDH. Finally, GMPLS can be deployed by organizations with mixed networks and services that require control of multiple technologies, including the optical domain with rapid bandwidth allocation as a key driver to GMPLS implementation. Some current issues with GMPLS include a lack of standards for interdomain routing, integration across nonGMPLS networks, and end-to-end instantiation.

In summary, MPLS technology offers service providers the capability to develop and deploy value-added services and to implement these services in an evolutionary manner. The service architecture is depicted in Figure 3-6.