Presentation 2019-03-08
[Encouragement Talk] Detecting Dependency between Modules of Deployment Automation Tool Using Reinforcement Learning Exploration Log
Manabu Nakanoya,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Under the evolution of the computing and the network virtualization technologies, deployment and configuration software has widely spread. Such software is being used to automate repeated routine work in the IT system operation, and has brought a big effect of improving service lead time and reducing system operation cost. However, the effect for atypical work such as ad-hoc healing and upgrade including application of patches is limited because the development cost of the workflow that is required to execute operation through the automation software is relatively high with respect to usage frequency. To use such software for atypical work more effectively, it is a promising approach to divide a workflow into reusable elements (modularization). In this approach, defining the dependencies between the reusable elements properly takes large amount of cost. In this work, we propose a dependency detection system using reinforcement learning exploration results obtained with the modularized elements (this is implemented as Ansible module.), and evaluate its effectiveness through the upgrade task of OpenStack that is an open source cloud infrastructure software.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Reinforcement Learning / Deployment Automation / Dependency Analysis / Ansible / OpenStack
Paper # ICM2018-70
Date of Issue 2019-02-28 (ICM)

Conference Information
Committee ICM
Conference Date 2019/3/7(2days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Kiyohito Yoshihara(KDDI Research)
Vice Chair Yoichi Yamashita(NTT-Neomait) / Takumi Miyoshi(Shibaura Inst. of Tech.)
Secretary Yoichi Yamashita(KDDI Research) / Takumi Miyoshi(NTT)
Assistant Yunchen Zhu(Hitachi)

Paper Information
Registration To Technical Committee on Information and Communication Management
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) [Encouragement Talk] Detecting Dependency between Modules of Deployment Automation Tool Using Reinforcement Learning Exploration Log
Sub Title (in English)
Keyword(1) Reinforcement Learning
Keyword(2) Deployment Automation
Keyword(3) Dependency Analysis
Keyword(4) Ansible
Keyword(5) OpenStack
1st Author's Name Manabu Nakanoya
1st Author's Affiliation NEC Corporation(NEC)
Date 2019-03-08
Paper # ICM2018-70
Volume (vol) vol.118
Number (no) ICM-483
Page pp.pp.117-122(ICM),
#Pages 6
Date of Issue 2019-02-28 (ICM)