Problem Statement


To understand why traffic engineering is needed in IP networks, let us first understand what happens in an IP network. As stated earlier, IP routers perform destination-based routing when sending traffic. This implies that the routers use a simple shortest path first algorithm to compute the shortest "distance" between themselves and the destination. This "distance" can be hop count, for protocols such as Routing Information Protocol (RIP), or least total metric (the sum of link metrics added along the path from the network element to the destination). It does not matter whether other alternate paths exist in the network. If there is traffic to send, the traffic always flows through the least cost path or shortest path first. Even if the traffic to be sent is more than the path can accommodate and the path itself is congested, the traffic is always sent on the shortest path. This results in traffic drop, as shown in Figure 8-1.

Figure 8-1. Shortest Path Routing


Assuming all links are OC-192 in Figure 8-1, if 8 Gbps of traffic is flowing between Seattle and New York, the network without traffic engineering can accommodate only 2 more Gbps of bandwidth between Seattle and New York on the shortest connection via Chicago. So, if an additional 4 Gbps of traffic now needs to be sent between Seattle and New York, this traffic flows along the Chicago-New York path on directly connected links. This results in a massive 2 Gbps of traffic drop. Even though other links do exist between Seattle and New York, such as via a Denver-Dallas path, these links are not used. In other words, even if other nonshortest paths that might have available link bandwidth exist in the network, because of shortest path first (SPF) computation, the traffic is always routed along the short path (the least distance). The nonshortest paths are not used; therefore, traffic drop occurs along the shortest path.

MPLS TE enables the operator to build a TE tunnel/LSP along the nonshortest path or a path that meets the bandwidth requirements (4 Gbps in the earlier example). In addition, MPLS TE maps traffic to this new path, thereby using links with available bandwidth. For example, in Figure 8-1, traffic going from Seattle to New York can now go through a TE tunnel built via Atlanta, Dallas, and Denver, as shown in Figure 8-2.

Figure 8-2. Shortest Path and Nonshortest Path Overview





MPLS and Next-Generation Networks(c) Foundations for NGN and Enterprise Virtualization
MPLS and Next-Generation Networks: Foundations for NGN and Enterprise Virtualization
ISBN: 1587201208
EAN: 2147483647
Year: 2006
Pages: 162

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net