خوارزمية موازنة الأحمال المقترحة في الحوسبة السحابية
No Thumbnail Available
Date
2017-05-24
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Cloud computing considers an important paradigm that provides
services and dynamic virtual resources to the consumers over the internet.
One of the main challenges and important fields for research in the cloud
computing environment is load balancing. So the main goal of this thesis is to
establish an effective load balancing algorithm for task scheduling.
In cloud computing environment, the demands are executed into the
virtual machines (VMs) and in the execution, it must be avoided a situation
some of the resources are under loaded states while there are resources
overloaded or heavy while achieving less the execution time of the tasks and
response time.
In this thesis, we present the Load Balancing Decision Algorithm (LBDA)
which aims to achieve load balancing for reducing the: Total completion time
(Makespan), Response time and Total execution time of tasks in cloud
computing environment.
The mechanism of LBDA is based on three functions, first: calculating the
virtual machine capacity, virtual machine load then determining VM states
(Under loaded, Balance, High balance, Overloaded) based on comparing VM
capacity, VM load and using three thresholds: threshold upper; threshold fair;
and threshold lower.
Second: estimating the time required to complete executing the task in
each VM. Finally, making a decision to distribute the tasks among the VMs
based on VM state and task time required. Where checked task time required
into VMs under states and selecting the fastest VM, when there are more than
one VMs have the same time to complete a task, LBDA assigns a task to VMs
II
which has the largest capacity. When all VMs become (Balance or High
balance) states, the tasks are distributed between VMs depending on the
fastest execution. The proposed LBDA was compared with Max-Min,
Shortest Job First and Round Robin algorithms. We applied many
experiments with various VMs’ capacity and: A different number of tasks on
a different number of VMs, The same number of tasks on a different number
of VMs, and A different number of tasks on a different number of VMs.
The performance evaluation results implemented in cloudsim toolkit 3.0.3
and the experiments demonstrate that our proposed LBDA outperformed to
other algorithms by reducing the completion time of tasks, response time, and
total execution time of tasks.