A Reinforcement Learning Based System for Minimizing Cloud Storage - - PowerPoint PPT Presentation

▶
a reinforcement learning based system for minimizing
SMART_READER_LITE
LIVE PREVIEW

A Reinforcement Learning Based System for Minimizing Cloud Storage - - PowerPoint PPT Presentation

Sep 24, 2022 176 likes •360 views

A Reinforcement Learning Based System for Minimizing Cloud Storage Service Cost Haoyu Wang 1 , Haiying Shen 1 , Qi Liu 1 , Kevin Zheng 1 , and Jie Xu 2 1 University of Virginia and 2 George Mason University ICPP2020 Online presentation Web

slide-1
SLIDE 1

A Reinforcement Learning Based System for Minimizing Cloud Storage Service Cost

Haoyu Wang1, Haiying Shen1, Qi Liu1, Kevin Zheng1, and Jie Xu2

1University of Virginia and 2George Mason University

ICPP2020 Online presentation

slide-2
SLIDE 2

Web application:

slide-3
SLIDE 3

Cloud Storage

Hot Cold

CSP: Cloud Service Provider Web Application User

Type of storage Archive

slide-4
SLIDE 4

US West Hot $0.0055 Cold $0.01 US East Hot $0.005 Cold $0.01

slide-5
SLIDE 5

Outline

  • How to minimize storage monetary cost
  • Related work
  • Wikipedia trace analysis
  • Markov decision process problem formulation
  • Main design
  • Performance evaluation
  • Conclusion
slide-6
SLIDE 6

Minimize storage monetary cost Different price is determined by:

  • Storage type
  • Read/write operation frequencies
  • Storage period
slide-7
SLIDE 7

Related work

  • Cloud storage payment minimization
  • Cloud resource pricing
  • Combining cloud providers

Unlike the above methods, the goal of our method is to minimize the total payment a cloud storage service customer made to a CSP by leveraging the different types of storage provided by the CSP.

slide-8
SLIDE 8

Trace analysis

slide-9
SLIDE 9

Trace analysis

slide-10
SLIDE 10

Trace analysis

slide-11
SLIDE 11

Problem formulation

Markov Decision Process M=(S,A,P,R) State space: Action space: Reward:

slide-12
SLIDE 12

Main design

  • 1. A3C algorithm used in

MiniCost

slide-13
SLIDE 13

Main design

  • 2. Concurrent requested

data files aggregation

slide-14
SLIDE 14

Performance evaluation

slide-15
SLIDE 15

Performance evaluation

slide-16
SLIDE 16

Conclusion

  • Analysis on the Wikipedia trace demonstrates that the substantial

request frequency variabilities may make it cost-inefficient for cloud storage service customer.

  • An RL based data storage types assignment algorithm that generates

data storage types assignment plans periodically can save monetary cost significantly.

  • Trace-driven experiment shows that our online RL based method can

achieve significant cost savings.

slide-17
SLIDE 17

Thank you! Questions?