The Need for Mpls Te

Routing in IP networks is governed by the need to get the traffic across the network as quickly as possible. That is why IP routing is based on the least-cost routing principle. Every IP routing protocol has a cost associated with the links in the networks. The accumulation of the cost of every link of a path is used to calculate the smallest cost path to forward traffic through the network. That cost is a single metric that is assigned to a link (for instance, Open Shortest Path First [OSPF] and Intermediate System-to-Intermediate System [IS-IS]), a composite metric (for instance, Interior Gateway Routing Protocol [IGRP] and Enhanced Interior Gateway Routing Protocol [EIGRP]), or simply a hop count (for instance, Routing Information Protocol [RIP] and RIP version 2).

The forwarding paradigm of IP is based on this least-cost path forwarding. Furthermore, IP packets are forwarded on every hop (router) based solely on the destination IP address and independently of the way the IP packets were forwarded on the routers before or after this hop. Also, the IP forwarding paradigm does not take into account the available bandwidth capacity of the link, which might differ significantly from the cost that is assigned to the link. Therefore, a router can keep forwarding IP traffic onto a link, even though that link is already dropping packets due to a lack of bandwidth to forward all the traffic flows for which the routing table points to that link. The result of this behavior in forwarding IP packets is that some links might be overutilized in the network, whereas other links might be underutilized. Of course, you can keep an eye on the traffic rates on the links of the network and plan an upgrade of the link capacity to accommodate the increased load. Adding bandwidth to the links is something that does not happen overnight; it needs planning and takes time. Because traffic patterns between sites can shift quite suddenly and are not always permanent, TE can bring a solution by steering the traffic or a portion of it away from the overloaded links. Look at Figure 8-1 for a sample network where IP forwarding is in place.

Figure 8-1 Network with IP Forwarding

If every link in this sample network has the same cost, the least cost path from router R1 to router R5 is the path R1-R2-R5. Clearly, all traffic from R1 to R5 will use the path R1-R2-R5, and the path R1-R3-R4-R5 will have no traffic. In a real network, things are not as black and white. Many traffic flows can exist, and the load on the links might vary greatly.

You can distribute the load more evenly by playing with the cost of the links for the particular routing protocol. That might distribute the traffic more evenly, but you can never distribute the load perfectly, because in real networks, the links are hardly ever of the same bandwidth capacity. In the network of Figure 8-1, you can ensure that the two paths look equal by making sure that the sum of the costs of the links in the path R1-R2-R5 and the path R1-R3-R4-R5 are equal. The result will be the load balancing of traffic between R1 and R5 on the two paths. This will be fine for the traffic between R1 and R5, but you will surely have traffic entering the network on R2 and leaving it on R4, and so on. This is the same problem, because two paths exist from R2 to R4; one path is two hops, and the other is three. You can have the same problem between the routers R3 and R5 or any of the others. In other words, the problem of loading the links equally with traffic is an impossible task if you just try to adjust the cost of each link in the network.

To further increase the complexity of the problem, at any given day, the speed of any of the links might be upgraded, allowing for more bandwidth on certain links. At that point, you need to plan again from scratch and change the cost of the links manually throughout the network. This is clearly not sustainable from an operational standpoint. MPLS TE is a solution for this problem in the following ways:

■ MPLS TE provides efficient spreading of traffic throughout the network, avoiding underutilized and overutilized links.

■ MPLS TE takes into account the configured (static) bandwidth of links.

■ MPLS TE takes link attributes into account (for instance, delay, jitter).

■ MPLS TE adapts automatically to changing bandwidth and link attributes.

■ Source-based routing is applied to the traffic-engineered load as opposed to IP destination-based routing.

MPLS TE allows for a TE scheme where the head end router of a label switched path (LSP) can calculate the most efficient route through the network toward the tail end router of the LSP. The head end router can do that if it has the topology of the network. Furthermore, the head end router needs to know the remaining bandwidth on all the links of the network. Finally, you need to enable MPLS on the routers so that you can establish LSPs end to end. The fact that label switching is used and not IP forwarding allows for source-based routing instead of IP destination-based routing. That is because MPLS does forwarding in the data plane by matching an incoming label in the label forwarding information base (LFIB) and swapping it with an outgoing label. Therefore, it is the head end label switching router (LSR) of the LSP that can determine the routing of the labeled packet, after all LSRs agree which labels to use for which LSP. Figure 8-2 shows an example of this source-based routing ability of MPLS TE.

Figure 8-2 MPLS TE Head End Router

Figure 8-2 MPLS TE Head End Router


To illustrate this concept, routers R6 and R7 have been added in front of router R1. Assume that routers R6 and R7 want to send traffic to R5. If this network is running IP forwarding only, this traffic follows the path R1-R2-R5 only, no matter what you configured on routers R6 and R7. That is because the forwarding of IP packets is done independently on every hop in the network. Therefore, router R1 does not know what routers R6 and R7 are up to, and it forwards the traffic according to its own forwarding decision based on the IP routing table. R6 and R7 might have different policies, though. R6 might want to send the traffic along the path R6-R1-R2-R5, whereas R7 might want to forward the traffic along the path R7-R1-R3-R4-R5, which is impossible to achieve in a plain IP network. If the network is running MPLS, you can set up these two paths as two different LSPs so that different labels are used. At router R1, the different incoming label value indicates whether the packet belongs to the LSP with R6 as the head end or the LSP with R7 as the head end. R1 then forwards the packet on one of the two LSPs, but it does not forward the packet according to its own will as is the case with plain IP forwarding.

You can deploy MPLS TE in any network that has LSRs. However, because the bandwidth and other attributes of the links have to be known by the head end LSR of the LSPs, the routing protocol used between MPLS TE endpoints (head end and tail end LSRs) has to be a link state routing protocol. With a link state routing protocol, each router builds a state of its own links, which is then flooded to all the other routers in the same area. This means that all routers in the area have all topology information of that area. The head end LSR can thus figure out how to lay out the MPLS traffic-engineered LSP. This allows for source-based routing. This LSP is called an MPLS TE tunnel. It is not like a GRE tunnel, however. A TE tunnel is unidirectional, because an LSP is unidirectional, and it has the TE tunnel configuration only on the head end LSR and not on the tail end LSR of the LSP. Furthermore, a TE tunnel must be signaled, whereas a GRE tunnel does not have to be.

If TE is enabled in the network, you can use it in two distinct ways. First, you can create MPLS TE tunnels between each pair of edge LSRs in your network. As such, you can steer all traffic in the network, avoid congestion in it, and give all traffic the characteristics (bandwidth, delay, jitter, and so on) it needs. A good example is MPLS VPN, where you can create one TE tunnel from every PE router to every other PE router. Second, you can enable MPLS TE everywhere in the network but not have TE tunnels until they are needed. You can create the TE tunnels on demand. A good example of this is when you create TE tunnels to steer traffic around a hotspot or overloaded point in the network. Equally important as how to steer traffic with the TE LSPs through the network is how to map traffic onto them. This is explained in the section "Forwarding Traffic onto MPLS TE Tunnels." When no traffic enters the TE tunnels, the TE tunnels are idle.

Micro Expression Master

Micro Expression Master

If You Could Read Everyone Life A Book You Can Have Better Career, Great Relationships And Become Successful. This Book Is One Of The Most Valuable Resources In The World When It Comes To Reading the smallest and tiniest body Language and know what people are thinking about.

Get My Free Ebook

Post a comment