Network Contention-Aware Cluster Scheduling with Reinforcement Learning

TL;DR

This paper introduces a reinforcement learning-based scheduling policy for GPU clusters that reduces network contention, improving overall training throughput and job completion times for distributed deep learning workloads.

Contribution

It formulates cluster scheduling as a reinforcement learning problem to create a network contention-aware policy that adapts dynamically, outperforming traditional policies.

Findings

Reduces average job completion time by up to 18.2%

Cuts tail job completion time by up to 20.7%

Balances job completion time with resource utilization effectively

Abstract

With continuous advances in deep learning, distributed training is becoming common in GPU clusters. Specifically, for emerging workloads with diverse amounts, ratios, and patterns of communication, we observe that network contention can significantly degrade training throughput. However, widely used scheduling policies often face limitations as they are agnostic to network contention between jobs. In this paper, we present a new approach to mitigate network contention in GPU clusters using reinforcement learning. We formulate GPU cluster scheduling as a reinforcement learning problem and opt to learn a network contention-aware scheduling policy that efficiently captures contention sensitivities and dynamically adapts scheduling decisions through continuous evaluation and improvement. We show that compared to widely used scheduling policies, our approach reduces average job completion time by up to 18.2\% and effectively cuts the tail job completion time by up to 20.7\% while allowing a preferable trade-off between average job completion time and resource utilization.