Cisco Cisco IOS Software Release 12.0(8)S 白書

ページ / 88
 MPLS Traffic Engineering 1
MPLS Traffic Engineering
Feature Overview
Multiprotocol Label Switching (MPLS) traffic engineering software enables an MPLS backbone to 
replicate and expand upon the traffic engineering capabilities of Layer 2 ATM and Frame Relay 
networks. 
Traffic engineering is essential for service provider and Internet service provider (ISP) backbones. 
Such backbones must support a high use of transmission capacity, and the networks must be very 
resilient, so that they can withstand link or node failures. 
MPLS traffic engineering provides an integrated approach to traffic engineering. With MPLS, traffic 
engineering capabilities are integrated into Layer 3, which optimizes the routing of IP traffic, given 
the constraints imposed by backbone capacity and topology.
MPLS traffic engineering:
Enhances standard IGPs, such as IS-IS or OSPF, to automatically map packets onto the 
appropriate traffic flows.
Transports traffic flows across a network using MPLS forwarding.
Determines the routes for traffic flows across a network based on the resources the traffic flow 
requires and the resources available in the network.
Employs "constraint-based routing," in which the path for a traffic flow is the shortest path that 
meets the resource requirements (constraints) of the traffic flow. In MPLS traffic engineering, the 
traffic flow has bandwidth requirements, media requirements, a priority versus other flows, and 
so on.
Recovers to link or node failures that change the topology of the backbone by adapting to a new 
set of constraints.
Why Use MPLS Traffic Engineering?
WAN connections are an expensive item in an ISP budget. Traffic engineering enables ISPs to route 
network traffic to offer the best service to their users in terms of throughput and delay. By making 
the service provider more efficient, traffic engineering reduces the cost of the network.
Currently, some ISPs base their services on an overlay model. In the overlay model, transmission 
facilities are managed by Layer 2 switching. The routers see only a fully meshed virtual topology, 
making most destinations appear one hop away. If you use the explicit Layer 2 transit layer, you can 
precisely control the ways in which traffic uses available bandwidth. However, the overlay model