SLAQ: Quality-Driven Scheduling for Distributed Machine Learning

Abstract

Training machine learning (ML) models with large datasets can incur significant resource contention on shared clusters. This training typically involves many iterations that continually improve the quality of the model. Yet in exploratory settings, better models can be obtained faster by directing resources to jobs with the most potential for improvement. We describe SLAQ, a cluster scheduling system for approximate ML training jobs that aims to maximize the overall job quality. When allocating cluster resources, SLAQ explores the quality-runtime trade-offs across multiple jobs to maximize system-wide quality improvement. To do so, SLAQ leverages the iterative nature of ML training algorithms, by collecting quality and resource usage information from concurrent jobs, and then generating highly-tailored quality-improvement predictions for future iterations. Experiments show that SLAQ achieves an average quality improvement of up to 73% and an average delay reduction of up to 44% on a large set of ML training jobs, compared to resource fairness schedulers.

arXiv commentAppeared in the 1st SysML Conference. Full paper published in ACM SoCC 2017

References in corpus (3)

• Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation
• Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding
• MLlib: Machine Learning in Apache Spark

Cited by in corpus (5)

• Characterizing Application Scheduling on Edge, Fog and Cloud Computing Resources
• Speculative Container Scheduling for Deep Learning Applications in a Kubernetes Cluster
• Semi-Dynamic Load Balancing: Efficient Distributed Learning in Non-Dedicated Environments
• Deep-Edge: An Efficient Framework for Deep Learning Model Update on Heterogeneous Edge
• Elastic CoCoA: Scaling In to Improve Convergence