Online Scheduling on Two Parallel Identical Machines with a Non-Clairvoyant Disruption

Abstract

This paper studies two parallel machine scheduling with a non-clairvoyantdisruption such that there happens one disruption on one machine at time 0 and its du-ration length can only be known at its end. As in [1], we introduce transportation timein the environment of disruption, aiming to minimize the maximum deviation of jobs'planned and actual completion times. We adopt online theory to describe the problemand focus on the case of unit length of job. We rst show that a greedily waiting strategyis 2-competitive. Our main result is a 3/2-competitive strategy BD which makes useof job transportation. We also prove a matching lower bound, implying that BD is anoptimal strategy.