3. Label distribution

We've seen how and on what principles MPLS packets are relayed, based on the label at the top of the label stack. What remains to be explained is how the LSRs in an MPLS cloud agree on the meaning to be attributed to the labels: this is known as label distribution. There are several ways of populating MPLS switching tables. The first is to do it manually, which is only realistic for a very limited number of equivalence classes (FEC). It is also possible to use a fully automatic protocol which, based on the information contained in the IP routing tables, constructs the LSPs for each of the equivalence classes recognized in the routing tables. With this approach, paths are constructed on a hop-by-hop basis, similar to that of IP routing protocols. An example of this approach is the new Label Distribution Protocol (LDP). The addition of label distribution information in BGP4 is another. Finally,...