Learning to Schedule DAG Tasks

TL;DR

This paper introduces a reinforcement learning-based method to improve DAG task scheduling by simplifying the problem and enhancing existing heuristics, leading to better performance on benchmark datasets.

Contribution

A novel learning-based approach that iteratively modifies DAGs to improve scheduling efficiency, compatible with existing heuristics like SJF and CP.

Findings

Significant performance improvements over traditional heuristics.

Consistent best results across various settings.

Effective reduction of DAG complexity for scheduling.

Abstract

Scheduling computational tasks represented by directed acyclic graphs (DAGs) is challenging because of its complexity. Conventional scheduling algorithms rely heavily on simple heuristics such as shortest job first (SJF) and critical path (CP), and are often lacking in scheduling quality. In this paper, we present a novel learning-based approach to scheduling DAG tasks. The algorithm employs a reinforcement learning agent to iteratively add directed edges to the DAG, one at a time, to enforce ordering (i.e., priorities of execution and resource allocation) of "tricky" job nodes. By doing so, the original DAG scheduling problem is dramatically reduced to a much simpler proxy problem, on which heuristic scheduling algorithms such as SJF and CP can be efficiently improved. Our approach can be easily applied to any existing heuristic scheduling algorithms. On the benchmark dataset of TPC-H, we show that our learning based approach can significantly improve over popular heuristic algorithms and consistently achieves the best performance among several methods under a variety of settings.