Resolving a question about span restoration: Do loopbacks involve a capacity penalty?

Under span restoration or “link protection,” situations arise where the end-to-end route of a restored path contains a “loopback,” wherein a path doubles back on itself over part of its route following a restoration action. Investigators often independently observe this and presume that loopback detection and avoidance would thus lead to a more capacity-efficient form of span restoration. Intuitively this may be compelling and a seemingly obvious conclusion. We show, however, that this is an oversimplified view and that loopback elimination would provide none or only trivial amounts of spare capacity savings. Theoretical reasoning and experimental validations based on forcer analysis are given confirming and explaining the findings. The main insight is that while spare capacity may be used in loopbacks for one or more failure scenarios, in an efficient design, there is always one or more other failure scenarios which do use all the spare capacity without loopbacks. There is then no savings possible from loopback elimination. The benefit to the field is that for the first time a quantitative assessment of this previously only qualitatively observed effect is provided and a theoretical framework is given within which we can understand in general when loopbacks will be associated with a possible capacity penalty and when there will be no such penalty at all. This adds to our basic knowledge in the area of survivable networking science.