Resource allocation in cloud virtual machines based on empirical service traces

SUMMARYOne of the key technologies in cloud computing is virtualization. Using virtualization, a system can optimize usage of resources, simplify management of infrastructure and software, and reduce hardware requirements. This research focuses on infrastructure as a service, resource allocation by...

Full description

Saved in:
Bibliographic Details
Published in:International journal of communication systems 2014-12, Vol.27 (12), p.4210-4225
Main Authors: Lin, Ching-Huang, Lu, Chien-Tung, Chen, Ying-Hsien, Li, Jung-Shian
Format: Article
Language:English
Subjects:
Central processing units
Cloud computing
Computer simulation
empirical service traces
Infrastructure
Mathematical models
resource allocation
resource model
Utilization
Virtual environments
virtual machine
Workload
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:SUMMARYOne of the key technologies in cloud computing is virtualization. Using virtualization, a system can optimize usage of resources, simplify management of infrastructure and software, and reduce hardware requirements. This research focuses on infrastructure as a service, resource allocation by providers for consumers, and explores the optimization of system utilization based on actual service traces of a real world cloud computing site. Before activating additional virtual machines (VM) for applications, the system examines CPU usage in the resource pools. The behavior of each VM can be estimated by monitoring the CPU usage for different types of services, and consequently, additional resources added or idle resources released. Based on historical observations of the required resources for each kind of service, the system can efficiently dispatch VMs. The proposed scheme can efficiently and effectively distribute resources to VMs for maximizing utilization of the cloud computing center. Copyright © 2013 John Wiley & Sons, Ltd. The allocation problem was simulated with Our CPU models. Results show the deviation in service level agreement (SLA) of unit virtual machines (VMs) with low workload is larger than the high workload VM. Even when the idle resources are released, some overloading situations may occur when allocating small reservations. However, through the verification of the simulations, we proof the CPU module applied for infrastructure as a service managers to adjust VM usage and efficiently dispatch resources using SLA and reservation parameters.
ISSN:1074-5351
1099-1131
DOI:10.1002/dac.2607