A Survey on Approximation Algorithms for Scheduling with Machine Unavailability

In this chapter we present recent contributions in the field of sequential job scheduling on network machines which work in parallel; these are subject to temporary unavailability. This unavailability can be either unforeseeable (online models) or known a priori (offline models). For the online models we are mainly interested in preemptive schedules for problem formulations where the machine unavailability is given by a probabilistic model; objectives of interest here are the sum of completion times and the makespan. Here, the non-preemptive case is essentially intractable. For the offline models we are interested in non-preemptive schedules where we consider the makespan objective; we present approximation algorithms which are complemented by suitable inapproximability results. Here, the preemptive model is polynomial-time solvable for large classes of settings.