Running a Jbo on a Collection of Dynamic Machines, with On-Line Restarts.211 Software Engineering

摘要

We consider the problem of running a job on a selected machine of a collection of211u001emachines. Each of these machines may become temporarily unavailable (busy) 211u001ewithout warning, in which case the scheduler is allowed to restart the job on a 211u001edifferent machine. The behavior of machines is characterized by a Markov chain, 211u001eand objective is to minimize completion time of the job. For several types of 211u001eMarkov chains, we present optimal policies.