Summary

International Conference on Emerging Technologies for Communications

2022

Session Number:S8

Session:

Number:S8-5

An Implementation of Effective Server Resource Management Scheme Using Deep Reinforcement Learning

Toshiki Kawakita,  Kimihiro Mizutani,  Satoru Kobayashi,  Kensuke Fukuda,  Osamu Akashi,  

pp.-

Publication Date:2022/11/29

Online ISSN:2188-5079

DOI:10.34385/proc.72.S8-5

PDF download (571.7KB)

Summary:
A container virtualization technology has been expanding rapidly in recent years, which can construct multiple application execution environments on a server. However, the server computing resources are shared among containers, and the performance of the containers may be degraded if appropriate computing resources are not allocated in busy case. In this study, we implement an automatic resources allocation scheme using reinforcement learning on consideration of computation load of containers to improve the performance degradation of the running applications. In addition, we evaluate it on the real container virtualization environment (i.e., Docker Environment) while a past literature evaluated it by simulation.