A survivable network is typically designed as 100% single failure restorable with
dual failure as the dominant failure scenarios. It is therefore of interest to assess
how strong a survivable network is under dual failure scenarios, which is referred
to as dual failure availability. Three indicators are currently employed as a surrogate
for dual failure availability, including restorability, service path unavailability,
and number of non-restored working capacities. It is common practice to enhance restorability
and lower service path unavailability by minimizing number of non-restored working
capacities. These three indicators, however, cannot evaluate availability accurately
because they ignore the inner connection of channels on either failed spans under
a specific dual failure scenario. Conversely, number of lost paths is proposed to
replace them in this paper. Theoretical analysis proves that number of lost paths
is more accurate compared with the traditional indicators. A case study further supports
the assertion that the new indicator is more accurate than the old ones in span-restorable
networks intuitively.
