CIRCUMFLEX: A Scheduling Optimizer for MapReduce Workloads Involving Shared Scans

We consider MapReduce clusters designed to support multiple concurrent jobs, concentrating on environments in which the number of distinct datasets is modest relative to the number of jobs. Many datasets in such scenarios will wind up being scanned by multiple concurrent Map phase jobs. As has been noticed previously, this scenario provides an opportunity for Map phase jobs to cooperate, sharing the scans of these datasets, and thus reducing the costs of such scans. Our paper has two main contributions. First, we present a new, novel and highly general method for sharing scans and thus amortizing their costs. This concept, which we will call cyclic piggybacking, is an alternative to the more traditional batching scheme described in the literature, and seems to have a number of advantages over that scheme. Second, we describe a method for optimizing schedules within the context of this cyclic piggybacking paradigm. This scheme, a significant but natural generalization of the recently introduced flex scheduler for MapReduce, can optimize a wide variety of metrics. Such cost functions include average response time, average stretch, and any minimax-type metric. The overall approach, including both cyclic piggybacking and the flex generalization, is called circumflex. We demonstrate the excellent performance of circumflex via a variety of simulation experiments, and we describe a practical implementation strategy.