In IP/optical multilayer designs, what is the relationship between links and spans?  Let's consider an IP network over a lower-layer network of fiber spans.  A span is the fiber connection between two adjacent amplifiers or optical nodes, but in network planning we sometimes use "span" to refer to the connection between two adjacent add/drop optical nodes (normally called an optical link or sublink).  The latter use of the term "span" is useful when we want to reserve "link" for bandwidth connectivity between IP routers.

Direct connectivity between two routers is an edge;  the implementation of this connectivity between an interface on a router on each end of the edge is called a link.  The set of links between the same pair of routers, traversing the same edge (as a sequence of fiber spans) is called a link bundle.  

A link (in an IP network) is routed over a sequence of spans, so a link in an IP network is a path in an optical network.  This means that a path mesh restoration scheme in the optical network may have similar properties to a link mesh restoration scheme (e.g. FRR) in an IP network, unless the connectivity of the optical network is much richer than that of the IP network.

No wonder why most papers just (try to make it simple and) assume that links, edges, and spans are synonymous.   As indicated above, this works fine for single-layer restoration but not so well for multilayer like IP/optical.