Work-in-Progress: Maximizing I/O throughput and Minimizing Performance Variation via Reinforcement Learning based I/O Merging for SSDs

Chao Wu*, Cheng Ji, Qiao Li, Chenchen Fu, Chun Jason Xue

*Corresponding author for this work

Research output: Chapters, Conference Papers, Creative and Literary WorksRGC 32 - Refereed conference paper (with host publication)peer-review

3 Citations (Scopus)

Abstract

Merge technique is widely adopted by I/O schedulers to maximize system I/O throughput. However merging operation could degrade the latency of individual I/O, thus incurring prolonged I/O latencies and enlarged I/O variations of I/O requests. In this case, the requirement of QoS (Quality of Service) performance will be violated. In order to improve QoS performance meanwhile providing high I/O throughput, this paper proposed a Reinforcement Learning based I/O merge approach. Through learning the characteristic of various I/O patterns, the proposed approach make merge decisions adaptive to different I/O workloads.
Original languageEnglish
Title of host publication2018 International Conference on Compilers, Architecture and Synthesis for Embedded Systems (CASES)
Subtitle of host publicationDigest of Technical Papers
PublisherIEEE
ISBN (Print)978-1-5386-5564-1
DOIs
Publication statusPublished - Sept 2018
Event2018 International Conference on Compilers, Architecture and Synthesis for Embedded Systems, CASES 2018 - Turin, Italy
Duration: 30 Sept 20185 Oct 2018

Publication series

Name2018 International Conference on Compilers, Architecture and Synthesis for Embedded Systems, CASES 2018

Conference

Conference2018 International Conference on Compilers, Architecture and Synthesis for Embedded Systems, CASES 2018
Country/TerritoryItaly
CityTurin
Period30/09/185/10/18

Research Keywords

  • I/O scheduler
  • Merge technique
  • performance variation
  • Reinforcement Learning
  • throughput
  • worst-case latency

Fingerprint

Dive into the research topics of 'Work-in-Progress: Maximizing I/O throughput and Minimizing Performance Variation via Reinforcement Learning based I/O Merging for SSDs'. Together they form a unique fingerprint.

Cite this