Algorithms for scheduling deadline-sensitive malleable tasks

Due to the ubiquity of batch data processing in cloud computing, the fundamental problem of scheduling malleable batch tasks and its extensions have received significant attention recently. In this paper, we consider an important model in which a set of n tasks is to be scheduled on C identical machines and each task is specified by a value, a workload, a deadline and a parallelism bound. Within the parallelism bound, the number of machines allocated to a task can vary over time without affecting its workload. For this model, we obtain two core results: a quantitative characterization of a sufficient and necessary condition such that a set of malleable batch tasks with deadlines can be scheduled on C machines, and a polynomial-time algorithm to produce such a feasible schedule. These core results provide a conceptual tool and an optimal scheduling algorithm that enable proposing new analyses and designs of algorithms and improving existing algorithms for extensive scheduling objectives.