Scheduling Inference Workloads on Distributed Edge Clusters with Reinforcement Learning

TL;DR

This paper presents ASET, a reinforcement learning-based scheduling algorithm designed for real-time inference workloads on distributed edge clusters, effectively adapting to network conditions and workloads to optimize performance.

Contribution

The paper introduces ASET, a novel RL-based scheduling policy for edge inference workloads that outperforms static policies in dynamic network environments.

Findings

ASET outperforms static scheduling policies in simulations.

Dynamic adaptation to network conditions improves inference latency.

Reinforcement learning effectively manages edge resource scheduling.

Abstract

Many real-time applications (e.g., Augmented/Virtual Reality, cognitive assistance) rely on Deep Neural Networks (DNNs) to process inference tasks. Edge computing is considered a key infrastructure to deploy such applications, as moving computation close to the data sources enables us to meet stringent latency and throughput requirements. However, the constrained nature of edge networks poses several additional challenges to the management of inference workloads: edge clusters can not provide unlimited processing power to DNN models, and often a trade-off between network and processing time should be considered when it comes to end-to-end delay requirements. In this paper, we focus on the problem of scheduling inference queries on DNN models in edge networks at short timescales (i.e., few milliseconds). By means of simulations, we analyze several policies in the realistic network settings and workloads of a large ISP, highlighting the need for a dynamic scheduling policy that can adapt to network conditions and workloads. We therefore design ASET, a Reinforcement Learning based scheduling algorithm able to adapt its decisions according to the system conditions. Our results show that ASET effectively provides the best performance compared to static policies when scheduling over a distributed pool of edge resources.