In the TCP/IP protocol suite, it is standard practice for the header corresponding to a given protocol layer to contain information that identifies the protocol used in the next higher layer. This information is needed so that the recipient of the PDU, when stripping off a given header, knows how to interpret the remaining bits so as to identify and process the header portion. For example, the IPv4 and IPv6 headers have a Protocol and Next Header field, respectively (Figure 14.5); TCP and UDP both have a Port field

(Figures 15.10 and 15.14), which can be used to identify the protocol on top of TCP or UDP. However, each MPLS node processes a packet whose top element is the MPLS label field, which contains no explicit information about the protocol that is encapsulated.

Typically, that protocol is IPv4, but it could be some other network-layer protocol.

a. Along an LSP, which MPLS nodes would need to identify the packet’s network layer protocol?

b. What conditions must we impose on the MPLS label in order for proper processing to occur?

c. Are such restrictions needed on all of the labels in an MPLS label stack or, if not, which ones?