Abstract
Nowadays, as the use of cloud computing service becomes more extensive and the customers welcome this service, an increasing trend in energy consumption and operational costs of these centers may be seen. To reduce operational costs, the providers should decrease energy consumption to an extent that Service Level Agreement (SLA) maintains at a desirable level. This paper adopts the virtual machine consolidation problem in cloud computing data centers as a solution to achieve this goal, putting forward solutions to make the decision regarding the necessity of migration from hosts and finding appropriate hosts as destinations of migration. Using time-series forecasting method and Double Exponential Smoothing (DES) technique, the proposed algorithm predicts CPU utilization in near future. It also proposes an optimal equation for the dynamic lower threshold. Comparing current and predicted CPU utilization with dynamic upper and lower thresholds, this algorithm identifies and categorizes underloaded and overloaded hosts. According to this categorization, migration then occurs from the hosts that meet the necessary conditions for migration. This paper identifies a certain type of hosts as “troublemaker hosts”. Most probably, the process of prediction and decision making regarding the necessity of migration will be disrupted in the case of these hosts. Upon encountering this type of hosts, the algorithm adopts policies to modify them or switch them to sleep mode, thereby preventing the adverse effects caused by their existence. The researchers excluded all overloaded, prone-to-be-overloaded, underloaded, and prone-to-be-underloaded hosts from the list of suitable hosts to find suitable hosts as destinations of migration. An average improvement of 86.2%, 28.4%, and 87.2% respectively for the number of migrations of virtual machines, energy consumption, and SLA violation is among the simulation achievements of this algorithm using Clouds tool.
Highlights
Cloud computing is a model that provides access to infrastructure including a set of configurable computing resources such as servers, storages, applications, services, etc
As mentioned in [12] and in light of the mechanism of finding underloaded hosts in [12], it is noteworthy that at load peak time, when the utilization of all hosts is at a high level, the hosts that have lower utilizations compared with other hosts will identify as underloaded hosts and the VMs thereon will migrate to other hosts
Research projects that require the work load of real data centres for simulation make use of the data pertaining to a 10-day workload from CoMon project [28], which is a monitoring infrastructure for PlanetLab and collected in March and April 2011. This data comprises CPU utilization data collected at 5-minute intervals from over thousands of operational VMs relating to service providers in more than 500 locations around the world
Summary
Cloud computing is a model that provides access to infrastructure including a set of configurable computing resources such as servers, storages, applications, services, etc. Failure to utilize them in a perfect manner will result in a huge energy loss [6] In this respect, Forrester research team observed that when a server is idle for 70% of the time, it consumes a power of almost 30% of the consumption peak power [7]; what mainly accounts for energy loss in cloud computing data centres is the use of equipment while their utilization is at low levels [6]. In this study, Proposed Algorithm makes decision based on dynamic upper and lower threshold as well as current and predicted CPU utilization. Decision making regarding the necessity of migration from hosts using the comparison of current and predicted CPU utilization with dynamic upper and lower thresholds as well as the identification and categorization of overloaded and underloaded hosts.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Advanced Computer Science and Applications
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
