International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 03 Issue: 04 | April-2016 www.irjet.net p-ISSN: 2395-0072
© 2016, IRJET ISO 9001:2008 Certified Journal Page 2734
A Review on Different Existing Load Balancing Approaches for Cloud
Saurabh Jain
1
, Dr. Varsha Sharma
2
1
Student, School of Information Technology, RGPV, Bhopal, MP, India
2
Assistant Professor, School of Information Technology, RGPV, Bhopal, MP, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Cloud computing is one of the emerging
technologies in modern IT era. It provides on demand IT
related resources to the client on the rent basis. The key
concept behind cloud computing is Dzvirtualizationdz.
Virtualization is an interface which separate hardware from
software and provides the benefits of server consolidation and
live migration. A main benefit of virtualization is that, it allows
creating multiple virtual machines (VMs) on a single physical
machine (PM). Since multiple users share the cloud resources,
so there may be situation when the PM is overloaded or under
loaded. VM migration is a unique solution to mitigate both
situations. Performance of the PM degrades when the PM is
overloaded or under loaded. If load balancing in cloud is not
handled properly, then it will increase the energy consumption
and minimize the provider profit. Hence, load balancing in
cloud is the prime requirement of the cloud provider. Load
balancing in cloud is a challenging task due to change in user
requirement at the run time. When the PM is overloaded or
underloaded, some VMs need to be migrated. VM migration
consists of three steps: source PM selection, VM selection and
the last step target PM selection. Numerous load balancing
algorithms have been proposed in the literature. This paper
discussed various existing load balancing approaches with
their comparisons.
Key Words: Virtualization, migration, energy efficient,
virtual machine, physical machine
1. INTRODUCTION
Cloud computing is one of the fascinating technology in the
field of computer science. It became so popular due to its
attractive features like on-demand services, utility based
model, ease of use etc. According to Gartner’s report [ͳ], the
Core technology behind the cloud computing is the
virtualization [2, 3]. Virtualization is the abstraction layer
between the hardware and software. It divides the physical
resources into the multiple types and allows the sharing of
physical resources. Virtualization increases the resource
utilization because multiple users share the same physical
resources but, it also introduces the need of load balancing. If
the load is distributed properly then it will increase the
resource utilization and reduce the energy consumption.
During the study of load balancing approaches it is found
that if the load is distributed effectively then it will minimize
the energy consumption and total simulation time. Load
balancing approach can be static or dynamic. Static approach
is more appropriate for the cloud. The one where
information about system is not important and working is
less complex comes under static scheme whereas in the
dynamic approach load balancing decision is taken based on
the previous system information.
Most of the existing load balancing [4-6] approaches use
thresholds to represent the overloaded and underutilized
situation where lower threshold represents the under
loaded situation and upper threshold is use to define the
overloaded situation. To handle the overloaded and under
loaded situations VM migration is used which move the VM
from one PM to another PM. VM migration is the important
feature of virtualization and is used to deal with the
overloaded, under loaded and hot spot situations. Hence, VM
migration allows the flexibility to resource provisioning.
Several load balancing approaches have been proposed in
the last few years. This paper gives the overview of some
approaches with their anomalies.
2. Following are the Goals of load balancing in
Cloud computing
1) To improve the performance of the cloud services.
2) To have a backup plan in case the system fails even
partially.
3) Increase the availability of the resources.
4) To maintain the system stability.
5) To minimize the number of SLA violations.
3. Challenging Issue in load balancing
Main challenging issues for the load balancing are [3]
1) The load balancing approach imposed some
overhead to process the request because it needs
some resources for the processing. So this overhead
should be minimize as possible
2) Since load in the cloud change dynamically, so it is
very difficult to design an efficient load balancing
approach for the cloud environment.
3) To balance the system, scheduler migrates the VM
which reduce the system performance. So a load
balancing approach must reduce the number of
migrations.