MPLS Traffic Engineering (MPLS TE) is a growing implementation in today's service provider networks. MPLS adoption in service provider networks has increased manifold due to its inherent TE capabilities. MPLS TE allows the MPLS-enabled network to replicate and expand upon the TE capabilities of Layer 2 ATM and Frame Relay networks. MPLS uses the reachability information provided by Layer 3 routing protocols and operates like a Layer 2 ATM network. With MPLS, TE capabilities are integrated into Layer 3, which can be implemented for efficient bandwidth utilization between routers in the SP network.
This chapter provides you with information on the operation and configuration of MPLS TE on Cisco routers.
TE is the process of steering traffic across to the backbone to facilitate efficient use of available bandwidth between a pair of routers. Prior to MPLS TE, traffic engineering was performed either by IP or by ATM, depending on the protocol in use between two edge routers in a network. Though the term "traffic engineering" has attained popularity and is used more in the context of MPLS TE today, traditional TE in IP networks was performed either by IP or by ATM.
TE with IP was mostly implemented by manipulation of interface cost when multiple paths existed between two endpoints in the network. In addition, static routes enabled traffic steering along a specific path to a destination. Figure 9-1 outlines a basic IP network with two customers, A and B, connected to the same service provider.
Figure 9-1 Traditional IP Networks
As illustrated in Figure 9-1, two paths exist between customer routers CE1-A and CE2-A via the provider network. If all links between the routers in Figure 9-1 were of equal cost, the preferred path between customer routers CE1-A and CE2-A would be the one with the minimum cost (via routers PE1-AS1, P3-AS1, and PE2-AS1) or PATH1. The same would apply for the customer routers CE1-B and CE2-B belonging to Customer B. If all the links were T3 links, for example, in the event of CE1-A sending 45 Mbps of traffic and CE1-B simultaneously sending 10 Mbps of traffic, some packets will be dropped at PE1-AS1 because the preferred path for both customers is using PATH1. The path PATH2 will not be utilized for traffic forwarding; therefore, TE can utilize this available bandwidth. To implement TE using IP whereby the paths PATH1 and PATH2 are either load balanced or used equally, we will need to implement IGP features such as maximum paths with variance or change the cost associated with the suboptimal path, PATH2, to make it equal to the current optimal path, PATH1. In an SP environment, this is often cumbersome to implement as the number of routers is much larger.
With ATM networks, the solution is a lot more feasible; PVCs can be configured between routers PE1-AS1 and PE2-AS1 with the same cost, but this would create a full mesh of PVCs between a group of routers. Implementing ATM for TE, however, has an inherent problem when a link or a node goes down. During link or node failure used in conjunction with ATM for TE, messages are flooded on the network. The Layer 3 topology must be predominantly fully meshed to take advantage of the Layer 2 TE implementation. Often, this might prove to be a scalability constraint for the IGP in use, due to issues with reconvergence at Layer 3.
The main advantage of implementing MPLS TE is that it provides a combination of ATM's TE capabilities along with the class of service (CoS) differentiation of IP. In MPLS TE, the headend router in the network controls the path taken by traffic to any particular destination in the network. The requirement to implement a full mesh of VCs, as in ATM, does not exist when implementing MPLS TE. Therefore, when MPLS TE is implemented, the IP network depicted in Figure 9-1 transforms into the label switched domain, as shown in Figure 9-2, in which the TE label switched paths or TE tunnels (Tunnel1 and Tunnel2) define paths that can be used by traffic between PE1-AS1 and PE2-AS1.
Figure 9-2 MPLS TE