Learning Scheduling Algorithms for Data Processing Clusters

TL;DR

This paper presents Decima, a machine learning-based scheduling system that automatically learns workload-specific algorithms for data processing clusters, significantly outperforming traditional heuristics in job completion times.

Contribution

Decima introduces a novel reinforcement learning approach with new representations and training methods tailored for complex, large-scale data processing scheduling.

Findings

Decima reduces average job completion time by at least 21%.

Achieves up to 2x speedup during high load periods.

Outperforms hand-tuned heuristics in real cluster tests.

Abstract

Efficiently scheduling data processing jobs on distributed compute clusters requires complex algorithms. Current systems, however, use simple generalized heuristics and ignore workload characteristics, since developing and tuning a scheduling policy for each workload is infeasible. In this paper, we show that modern machine learning techniques can generate highly-efficient policies automatically. Decima uses reinforcement learning (RL) and neural networks to learn workload-specific scheduling algorithms without any human instruction beyond a high-level objective such as minimizing average job completion time. Off-the-shelf RL techniques, however, cannot handle the complexity and scale of the scheduling problem. To build Decima, we had to develop new representations for jobs' dependency graphs, design scalable RL models, and invent RL training methods for dealing with continuous stochastic job arrivals. Our prototype integration with Spark on a 25-node cluster shows that Decima improves the average job completion time over hand-tuned scheduling heuristics by at least 21%, achieving up to 2x improvement during periods of high cluster load.