Publisher DOI: 10.1109/ATNAC.2018.8615319
Title: Towards Prediction of Power Consumption of Virtual Machines for Varying Loads
Language: English
Authors: Salam, Humaira Abdul 
Davoli, Franco 
Carrega, Alessandro 
Timm-Giel, Andreas  
Keywords: power modeling;power profiling;regression model;Virtual machine
Issue Date: 16-Jan-2019
Source: 2018 28th International Telecommunication Networks and Applications Conference, ITNAC 2018: 8615319 1-6 (2019-01-16)
Abstract (english): Power management and load balancing in data centers are becoming critical with the growing size of the infrastructure. In front of an increasing number of cloud networks, virtual machines (VMs) and containers, smart management and control decisions are required, in order to instantiate or mobilize these virtual components. As regards reducing power consumption, it is also essential to consolidate virtual resources on the minimum possible number of servers compatible with performance requirements. In such a dynamic scenario, estimating the power that can be ascribed to a specific virtual component and its relation with the offered workload could be beneficial for optimized resource scheduling. Predicting the power consumption caused by a specific virtual component is however challenging. In this paper, two power models that relate a VM workload with the fraction of power consumed attributable to the VM are developed based on power profiling of a server. The accuracy of the proposed models is improved approximately by 3% compared to other existing models. © 2018 IEEE.
Conference: 28th International Telecommunication Networks and Applications Conference 
ISBN: 978-153867177-1
Institute: Kommunikationsnetze E-4 
Type: InProceedings (Aufsatz / Paper einer Konferenz etc.)
Appears in Collections:Publications without fulltext

Show full item record

Page view(s)

Last Week
Last month
checked on Sep 27, 2020

Google ScholarTM


Add Files to Item

Note about this record


Items in TORE are protected by copyright, with all rights reserved, unless otherwise indicated.