arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2301.13618cs.LGcs.NIcs.SYeess.SY
收藏

基于强化学习的分布式边缘集群推理负载调度

Scheduling Inference Workloads on Distributed Edge Clusters with Reinforcement Learning

Politecnico di Torino(都灵理工大学);Saint Louis University(圣路易斯大学);Universitat Politècnica de Catalunya(加泰罗尼亚理工大学);Telefónica Research(西班牙电信研究院)

Gabriele Castellano, Juan-José Nieto, Jordi Luque, Ferrán Diego, Carlos Segura, Diego Perino, Flavio Esposito, Fulvio Risso, Aravindh Raman

摘要:

许多实时应用(例如,增强/虚拟现实、认知辅助)依赖于深度神经网络(DNN)来处理推理任务。边缘计算被认为是部署此类应用程序的关键基础设施,因为将计算移动到数据源附近使我们能够满足严格的延迟和吞吐量要求。然而,边缘网络的约束性质给推理工作负载的管理带来了一些额外的挑战:边缘集群无法为DNN模型提供无限的处理能力,并且在端到端延迟要求方面,通常应该考虑网络和处理时间之间的权衡。在本文中,我们专注于在短时间尺度(即,几毫秒)。通过模拟,我们分析了几个政策在现实的网络设置和工作负载的一个大型ISP,强调需要一个动态的调度策略,可以适应网络条件和工作负载。因此,我们设计了ASET,一个基于强化学习的调度算法,能够根据系统条件调整其决策。我们的研究结果表明,ASET有效地提供了最好的性能相比,静态策略调度时,在一个分布式的边缘资源池。

英文摘要:

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.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑