MPLS is part new technology, part throwback to older technologies. Its power is really in how it marries both new and old to get the best of both worlds.
Routing works by choosing the best next hop router-by-router across a network. Each packet is routed independently, and even if the packet in question is the second part of a fragment, the packet gets the whole router header examination and destination lookup treatment, even though 99 percent of the time, the packet is going to exactly the same place.
MPLS was originally used by service providers to address the situation shown in Figure 16-1. Why route each packet hop-by-hop, when it is immediately obvious at the first router on the left that the packet has to make its way to the router on the right.
Routing is effective for networks that are resilient and that have widely dispersed destinations, but switching is more effective for networks with stable paths and a set of limited destinations. Also, paths provide a convenient way to apply CoS parameters to a packet flow, because the path that packets follow should have consistent bandwidth and delay parameters.
To understand how MPLS leverages both new and old, you must have a firm grasp on the following topics: