OSDI文章分享，论文（LinnOS: Predictability on Unpredictable Flash Storage with a Light Neural Network）的结构化文摘

最新推荐文章于 2025-01-08 15:34:14 发布

结构化文摘

最新推荐文章于 2025-01-08 15:34:14 发布

阅读量331

点赞数

CC 4.0 BY-SA版权

分类专栏：一分钟看懂操作系统顶会论文OSDI 文章标签：操作系统神经网络储存器

本文链接：https://blog.youkuaiyun.com/bookmashup/article/details/118304792

一分钟看懂操作系统顶会论文OSDI 专栏收录该内容

34 篇文章 ¥9.90 ¥99.00

订阅专栏

超级会员免费看

LinnOS是一款操作系统，通过轻量级神经网络预测SSD性能，以每IO为基础，提升并行存储应用的性能可预测性。它支持黑盒设备和真实生产跟踪，无需用户额外输入，且优于传统机制。评估显示，与保守策略和基于启发式的方法相比，LinnOS平均I/O延迟降低9.6-79.6%，预测准确率高达87-97%，每个I/O的推理开销仅4-6μs，证明了在操作系统中集成机器学习进行实时决策的可行性。

Authors:

Mingzhe Hao, Levent Toksoz, and Nanqinqin Li, University of Chicago; Edward Edberg Halim, Surya University; Henry Hoffmann and Haryadi S. Gunawi, University of Chicago

Abstract:

This paper presents LinnOS, an operating system that leverages a light neural network for inferring SSD performance at a very fine — per-IO — granularity and helps parallel storage applications achieve performance predictability. LinnOS supports black-box devices and real production traces without requiring any extra input from users, while outperforming industrial mechanisms and other approaches. Our evaluation shows that, compared to hedging and

了解本专栏