Chapter 5 - Traffic Engineering

Abstract

The Internet is a collection of nodes and links with the purpose of delivering IP datagrams from a source host to a destination host. The conventional routing protocols are based on “shortest path first” (SPF) routing, in which each datagram is routed on the shortest path between the source, and the destination. In source routing, each datagram contains a list of IP addresses of the links or nodes it must pass through, in order, on the way to its destination. The SPF routing is performed at each stage on the next address in the list so that the datagram may be sent along a path under the control of the source application. A controlled approach favored by many network operators is offline traffic engineering. In this mode of operation, all decisions about how to route traffic are made by one (or a few) centralized servers. Such servers can keep a coordinated view of the network and react to changes in traffic loads. They can supply source routes to hosts, tunneling information to selected routers, or constraint-based routing instructions to all routers in the network. Traffic engineering is increasingly utilizing Multiprotocol Label Switching (MPLS) to meet its tunneling needs. MPLS provides data tunnels through the network that can be placed according to explicit routes calculated offline or online and modified in real time.