Energy efficiency in cloud computing based on mixture power spectral density prediction

Dinh Mao Bui, Nguyen Anh Tu, Eui Nam Huh

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)

Abstract

Due to the budget and environmental issues, adaptive energy efficiency receives a lot of attention these days, especially for cloud computing. In the previous research, we developed a combined methodology based on nonparametric prediction and convex optimization to produce proactive energy efficiency-oriented solution. In this work, the predictive analysis was further enhanced by deriving the mixture power spectral density to model the complex cloud monitoring statistics. By engaging the improved technique to the predictive analysis, the prediction process was more adaptive to handle the fluctuation in system utilization. As a consequence, the optimization process could subsequently produce more appropriate setting for energy savings. After the infrastructure setting has been made available, the instruction of virtual machine migration was created and implemented by the cloud orchestrator. This instruction condensed the services into the pool of active facilities, satisfying the objective of power efficiency. Eventually, any physical machine out of the power configuration would be gradually terminated. Compared to our former method, the effectiveness of the proposed technique has been proven by cutting down 4.92% of energy consumption, while still maintaining a similar quality of services.

Original languageEnglish
Pages (from-to)2998-3023
Number of pages26
JournalJournal of Supercomputing
Volume77
Issue number3
DOIs
Publication statusPublished - Mar 2021

Keywords

  • Cloud computing
  • Energy efficiency
  • Power spectral density
  • Prediction
  • Virtual machines consolidation

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Software
  • Information Systems
  • Hardware and Architecture

Fingerprint

Dive into the research topics of 'Energy efficiency in cloud computing based on mixture power spectral density prediction'. Together they form a unique fingerprint.

Cite this