Journal of Physics: Conference Series

PAPER • OPEN ACCESS You may also like

- A global plan policy for coherent co-
Virtual Machine Allocation Policy for Load operation in distributed dynamic load
balancing algorithms
Balancing M Kara

- Model of load balancing using reliable

algorithm with multi-agent system
To cite this article: M. Shabina Ghafir et al 2021 J. Phys.: Conf. Ser. 2070 012129 M F Afriansyah, M Somantri and M A
Riyadi

- Load Balancing Scheduling of Power

Network in Cloud Computing Environment
Dongmei Bin, Tong Yu and Li Xin
View the article online for updates and enhancements.

This content was downloaded from IP address 206.127.220.231 on 18/11/2021 at 07:46

ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

Virtual Machine Allocation Policy for Load Balancing

Shabina Ghafir M., Afshar Alam, Farheen Siddiqui, Sameena Naaz

Jamia Hamdard New Delhi, India Jamia Hamdard New Delhi, India Jamia Hamdard, New
Delhi , India, Jamia Hamdard New Delhi, India
[email protected] [email protected] [email protected]
[email protected]

[email protected]

Abstract. This paper focuses on the VM allocation policies for load balancing in cloud
computing environment. Intermittent nature of balancing the load scheme into the cloud
computing becomes a challenging job and it also affects the load balancing of the cloud. The
suggested proposed model generates and step-up the VM allocation policies but also
transforms the generated cloud workload. Furthermore, to improve the workload distribution of
workload and stability of the overall cloud computing environment the load balancing
algorithm is most important for load balancing. The work of load balancing is equally effective
in the cloud computing environment and it is most essential one for load balancing algorithms
to take care of all issues at the time of the work load. The researchers studied different
algorithms to solve the problems of load balancing that generate problems during the
distribution of workloads. The analysis VM allocation policies are tested on CloudSim
environment and the results, and discussion is about to which one VM allocation policy is
superior. Keywords—Load Balancing, Grid-computing, Cloud Computing, VM allocation
policies, Fuzzy

1. Introduction

Cloud computing has become one of the famed fields of Computer science and because of its
flexibility and backing a wide range of users to available distributed resources/ infrastructure over the
Internet. Cloud computing has become a standout amongst most fascinating specialized fields in the
modern era. It has shown its effect on data stockpiling, information technology, programming
designing, and organizations. The National Institute of Standards and Technology (NIST)
characterizes cloud computing as : “the cloud computing is a paradigm to enable access for resources
pooling, convenient, on-demand, ubiquitous, which can be easily delivered with different types of
service provider interaction” [1].

Cloud computing is a new paradigm in field of computing and is considered as the fast growing new
advent, which is quickly developing at a greater pace and attracting increasingly new consumers and
vendors. The fast development of cloud computing is further expanded by the rising computing
innovations, which are being made at moderate and sensible expenses of infrastructure and capacity
capabilities. The term cloud computing implies the data are put away and go through Internet and not

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

from customary path from PC's hard drive. Cloud computing [2] has this inceptions in the days when
flowcharts and presentations, they speak to the servers infrastructure of the Internet. Local storage and
computing are the points at which a data are put away or projects are kept running from the hard drive,
thus enabling clients to have simple and quick access to data. The way that there is a committed
equipment server in a habitation, does not necessarily imply the presence of cloud computing. [3] The
information should be accessible over the Internet, or possibly the information should be synchronized
with data over the Internet. Cloud computing includes outsourcing of the computing assets with the
abilities of superfluous asset adaptability, on-request provisioning with moderate expenses.

Figure 1. CloudSim structure

The major advantage of cloud computing is the reduction of heavy investment costs in the IT
infrastructure. According to [4] and [5] the definition of cloud: “A Cloud is a sort of parallel and
conveyed framework comprising of an accumulation of interconnected and virtualized PCs that are
progressively provisioned and introduced as at least one brought together computing assets in view of
administration level understandings set up through arrangement between the administration provider
and clients.” Cloud computing considered to be highly beneficial in small businesses [6] since, it
admits innovations that weren't before in term of expenditures available for them as far as cash
spending; and it offers leverage for them by helping them with contending with other independent
ventures or even with huge ones. The cost of somebody coming and settling/introducing an application
is reduced and cash can be spared by the organization. Using software that are on the cloud is often
less costly than purchasing new ones. Thus, to utilize one multi-application cloud benefits everyone in
the organization.
The applications that live on the cloud can fully integrate into the organization in the light of the API
that finds the application that is good for the organizational objectives. The organization does not have
to burn cash for this because cloud computing is regularly overhauled. For organization, cloud
computing is a means of reducing the costs of the enterprise [7].
In Cloud Computing in order to guarantee economic benefits distributed among cloud customers [08]
scalable resources are provisioned with dynamism as a service. Various layers of cloud are operational

2
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

based on the type of service assured by the cloud server of which three layers from the basic core [8].
The first layer Infrastructure-as-a-Service (IaaS) consists of basic hardware requirements such as the
Memory and Storage resources. The most common example of IaaS is the Amazon Elastic Compute
Cloud (EC2). The next layer, second from the last is the Platform-as-a-Service (PaaS) which provides
the design and deployment of applications using Python, Java, etc. The best example of PaaS is the
Google App Engine.
The topmost layer namely Software-as-a-Service (SaaS) is the universally accepted and trustworthy
service that supports users to access the applications without the need for hardware or software
knowledge. Internet browsers and search engines are the fundamental tools required for accessing
cloud services.
As cloud can be accessible from any location anytime via primary commodity hardware its desire
is more or more day by day. Due to this more or more demand it requires to provide security to data as
well as high performance so that it can be advantageous for the users as well as for the service
providers. To reach the above purposes Load balancers are used in clouds. Load balancing is a way to
efficient manner distributes workload/traffic to the available nodes/resources over the network. It
optimizes the resource utilization by distributing work load effectually, reduces response time and
maximizes the throughput; it is also required for it to be fault tolerant and secure. Therefore load
balancing is one of vital parts of the cloud to enhance its performance. Cloud computing operates in a
distributed manner which comprises a large number of service nodes that performs several specific
tasks collectively in coordination with the other nodes. These systems and environments are highly
stable and also can handle requests from multiple users concurrently. But the explosive rise in data in
current trends attracts more service requests and hence the load on the cloud imposes a performance
challenge on the service provider. Various load balancing algorithms has been proposed for load
balancing, here we are try to solve issue of load balancing by using CloudSim and WorkflowSim. [9]
Figure -1 (CloudSim core simulation engine) showing the CloudSim structure that is three-layer
structure. In CloudSim simulator is the lower layer that is, known as simulation engine. The
functionality of this layer is to create real cloud environment and can't be modified by users. The lower
to upper layer is work as an interface between the cloudlets and VMs, VM allocation policy
algorithm.[16-21]

2. Challenges in Cloud Computing

From the beginning of the cloud computing history, certain challenges have remained unattended for
industries and the clients [10]. The most important issues are: interoperability between clouds,
generation of standards, security, fault tolerance and organizational aspects. Cloud computing has
generated a paradigm shift in the application and management of information technology
infrastructure, changing the enterprise and technology aspects of IT [11].

Yet, as with any significant change ever, there are both supporters and skeptics. Transferring IT to
cloud is a challenging process assignment which incorporates both authoritative and specialized
difficulties [12].
The cloud is different worldview which doesn't have a suitable one-sentence definition; it includes
numerous variables, and in this way change to a cloud-based process might confounding [13].
This multifaceted nature combined with instability throws up various hierarchical cloud-selection
barriers, which pose, specialized and authoritative difficulties. [14] Though the significance of
hierarchical difficulties is hard to measure; they are fundamental to the decision making process [15]
and therefore, need to be addressed.
Cloud computing is a service oriented model for providing IT infrastructures and applications.
Models and the way in which interoperability has taken account of the arrangements put forward by
distinctive merchants is a guide for achieving the objective. A key obstacle to the collection of cloud

3
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

figures is the way in which the vendors are safe. The customers are unable to turn to another contender
because the vendor is safe. In fact, as specified in the agreement, a fee should be paid and a measure of
time should be retained.

3. Deployment Models of Cloud Computing

The following four are the most commonly used types of cloud:
3.1. Public Cloud
A public cloud provides open access to its resources publically to every user [10]. It serves multiple
users in which the cloud providers can either set up their own infrastructure or utilize third party
infrastructure for allowing access to the resources. These resources are provided either free of cost or
on pay-per-usage. Public cloud is considered to the cost-efficient due to shared usage of resources and
some of the most popular cloud providers are Google, Amazon Ec2, etc.
3.2. Private Cloud
Private cloud is also identified as internal cloud employed in closed communities to provide higher
security and privacy with complete fulfillment of customer needs [11]. Private clouds enable business
organizations to utilize the cloud as the centralized access to its resources from various places
simultaneously, the different departments of the organization. Private cloud is installed either as on-
site or is out-sourced. On-site clouds are installed at customer locations while the out-sourced clouds
are installed in the host organizations. Some of the most common private cloud providers are VMware
Cloud Infrastructure Suite, Amazon VPC (Virtual Private Cloud), and Microsoft ECI datacenter.
3.3. Community Cloud
This cloud model is shared mutually among multiple companies, which have like-to-like policies and
norms. Organizations like banks and chits have related services such as privacy, security and
obligations and hence can be share the single clouds for cost efficiency. This model is also installed
either as on-site or out-sourced community clouds. Two most common examples of community clouds
are Google Apps for Government [9].
3.4. Hybrid Cloud
Hybrid cloud is the mixture of the private, public and community cloud which integrates various
benefits of all other deployment models such as cost efficiency, security, privacy, etc. of all the
deployment models [09]. This model is specifically used by organizations for withholding delicate
hidden information. However, due to limitations involved in taking care of requests, this model uses
open clou d for less sensitive data to deal with the loads.

4. Merits and Demerits of Cloud Computing

Cloud computing has a variety of advantages [23], some of which are discussed below:
4.1 Cost efficient
Since desktop programming is cost expensive for companies, it may be appropriate to pay charges for
various customers. However, cloud computing can be used at a lower price, thus reducing the
company's costs.
4.2 Abundant storage
The use of the cloud offers access to a vast amount of storage space for data.
4.3 Backup and recovery
It is much simpler to provide a backup and restore essential documents and save them on a physical
device.

4
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

4.4 Automatic software integration

The integration of software appears to take place promptly while using a cloud, which ensures that the
client has to endeavor to integrate the application into its system.
4.5 Easy access to information
The customer can access the data at any point it likes, the area or gadget it needs in any case, after the
customer has been enrolled in the cloud.
4.6 Faster deployment
After choosing the correct utility technique, all frameworks are prepared to work within minutes.
Though this environment has many advantages, there are some disadvantages of using cloud
computing paradigm [23], and they are discussed as follows:
4.7 Technical issues
Though cloud services can be accessed anytime from any location, there are certain times when the
systems have some dysfunctions, thus reducing its access.
4.8 Security issues
As mentioned in the previous sections, there are plenty of security problems due to non-availability of
strict regulations. Further the data in cloud are also prone to multiple attacks.
4.9 Performance issues
The issues such as load, task allocation, resource Utilization, energy management, etc., significantly
tend to minimize the efficiency of cloud paradigm.

5. CloudSimToolKit

CloudSim[16] is simulation software that is helping out the researcher for validate performance of their
propose algorithms result as real environment. This simulating tool is written in Java. This is Java
Library that is help out to creating the
simulation environment. In this simulation
environment easily researcher can add their
propose work and can evaluate the propose
work performance evaluate impact of
strategies from various performance
parameter, from cost/profit to speed up of
various VM allocation policy algorithms
execution time. This simulation tool is
having a Datacenter. The DataCenter
functionality is working as an IaaS provider.
Figure 1 (CloudSim core simulation engine)
showing the CloudSim structure that is three
layer structure. In CloudSim simulator is the
lower layer that is, known as simulation
engine. The functionality of this layer is to
create real cloud environment and can't be
modified by users. The lower to upper layer
is work as an interface between the cloudlets
Figure 2. CloudSim Classes
and VMs, VM allocation policy algorithm.

5
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

5.1. DataCenter
Datacenter working as on requests for VMs from brokers. In this process Datacenter create the VMs
in hosts.
5.2. Datacenter Broker Simple
This is simple java class CloudSim, this will responsible for only submits a list of VMs to be created,
schedules Cloudlets sequentially on them. When we want to implements the proposed scheduling
policy and/or policy for generation of VM requests and Cloudlets. The basic requirement is the need to
create own Broker.
Then to implement the VM allocation policies or algorithms, need to follows these steps:
1 First step is to need extend DatacenterSimple class
2. Afterward, for the purpose of describe periodic event need to define a new tag .
3. Next step is to override process Other Event. to This is regarding to find out the periodic event, and
there is a need to call handler for it4.Next step is for the handler and we use handler method
implement in cloud simulator for implement purpose. Finally, this will help to schedules the next call
for the event. Figure -1 showing the cloudsim engine, this image is showing how to simulator working
Figure -2 showing the class diagram, according to figure-2 in CloudSim having total of 12 packages
that has containing several classes. The most important package out of 12 packages is
org.cloudbus.cloudsim. The cloudsim class having codes for modeling variant cloud entities. In the
cloud entities are a Datacenter, a Host, a Cloudlet and VMs. In this package we can overwrite or
extend these classes to define the new cloud computing algorithms, This package will play a vital role
for validate the new propose algorithm performance.[22-24]
The CloudSim package further divided into two categories
1) Core entity classes
• Cloudlet
• Host,
• Datacenter
• VM
2) Associated classes.
• VmScheduler
• VmAllocationPolicy
• CloudletSched-uler
• UtilizationModel,
• DatacenterBroker
A Datacenter, a Host, a Cloudlet and VMs .In this package we can overwrite or extend these classes to
define the new cloud computing algorithms, this package is play a vital role for validate the new
propose algorithm performance.[25]

6. Virtual Machine Allocation

In cloud computing field, particularly within cloud allocation area, virtual machine provisioning can
be very challenging. A client on the host machine demands every virtual machine assigned to a
compliant host. This implies that the virtual machine relies on the resources of the real machine to
operate the virtual machine, such as storage and memory (Ezugwu, Buhari and Junaidu, 2013).

6
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

It can go through multiple stages when a virtual machine is allocated. The first step in the process is
the client/end-user request, including details like the virtual hardware specifications, operating system,
and amount of time needed for VM (Virtual Machine), including a SLA (The customer and host
service level agreement is decided during the first stage. Second, the application must go through
automatic or manual phase duration. The request is finally executed. This part of provisioning
provides access to the VM to the client.

As the cloud provider must have the ability to handle this when a client submits a request for a VM,
whether the order does not need to be rejected, this may be passed on to a cloud broker while
necessary resources outweigh cloud provider's capacity.

The SLA is kind of deal between client and cloud provider; there are service level goals (SLO) under
the Service Level Agreement; these are the metrics of performance that cloud provider must follow in
order not to violate the SLA between customer and cloud provider. Bandwidth, network strength, and
response time include all examples of SLOs.

Usually, if SLA is formal, it would be a formal document written in contractual agreement that will
have each SLO specification. Hence SLA can be either informal or formal between cloud provider and
client. For a violation between the cloud provider and the client, the document may also have a penalty
or fines in negotiated SLA. Nevertheless the SLA for which there are no SLA records available could
be informal. In the service given to them, consumers would expect a certain degree of quality,
regardless of whether the SLA is formal or informal. If cloud provider keeps failing to satisfy SLA
agreement, a provider may face harm to company's credibility, resulting in the loss of profit for cloud
provider.

Biggest care is taken to help reduce power usage when it comes to VM role of cloud provider,
because if the specification does not require 100% host utilization, spinning up a host for each order,
several VMs can be run on the same host, another priority is to provide more than one data centre in
one place or around the world to the cloud provider. Because if the client wanted low latency, placing
the VM in a centre far away from the client might have a significant impact on the experience of the
end user and potential violation of the SLA.

7. Allocation OF Virtual Machine: Architecture

Architecture consists of several mechanisms under virtual machines that can be partitioned into two
major areas:

Front End- This consists of software, such as a web browser like Internet Explorer or similar, that
relates to the customer/client face that is necessary to access the cloud.

Back End: This includes a large data storage facility, security system, services, models for
implementation, rules, monitoring, virtual machines, and cloud servers themselves.

7.1. Cloud Computing Monitoring

In order to optimize performance and minimize overhead costs, close monitoring is needed when
implementing virtual machines, because getting hosts running idle without assigned tasks will be a
wastage of power, there are several kinds of software that can control virtual machines like Paessler,
solarWinds and Veeam One (PC & Network Download-www.PCWDLD.com, 2018).

7
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

And with monitoring, if a host is overused, the virtual machines may be migrated to another host to
ease CPU, storage resources and memory, in addition, if hosts experience anyone potential hardware
malfunction, all virtual machines might be moved to other host by using migration and necessary
hardware/software is placed, With no need for power down virtual machine for concluding migration,
a live migration is feasible, or if there are restricted resources, virtual machine might have been moved
as a temporary step to another cloud provider, but if that was to migrate to a various cloud provider, it
might have a possible effect on the client/customer SLA.

7.2. Features of Front and Backend

The cloud provider handles features such as networking, storage, servers, virtualization and protection,
and features such as security are extremely essential as several virtual machines can be run on the
same host, as only those individuals need to guarantee the correct access rights.

8. VM Allocation Policies Algorithm

In this section our focus on CloudSim’s VM allocation Policies and doing simulation in CloudSim
tool. VM allocation is spin-off/provision of a VM on a physical host in which a certain predetermined
memory, CPU/vCPU, VLAN, storage, IP address, etc. may be allocated to the VM as part of the cloud
initiative or as part of the virtualization initiative. The following may be more general allocation of
cloud computing resources: Allocate additional private cloud resources to a VM (vertical scale-up)
(CPU, memory, storage, network, VFW, VLB, etc.) Allocate additional private cloud VMs where
VMs are considered a resource (horizontal scale-out) as part of the bursting of workload, allocate
additional external public cloud/IaaS resources DBaaS Note: Resource allocation for cloud computing
is typically automated for faster processing time and minimum human error by its orchestrator. Figure-
2 is describing the flow of VM allocation policies algorithm. Our algorithms are performance measure
by memory utilization, CPU utilization and bandwidth utilization, as these all are playing most
important role for cloud computing service faster performance[26].
8.1. Simple VM allocation policy
This is a simple VM allocation policy and these results are standard. This is default policy in
cloudsim environment.
8.2. Round Robin VM allocation policy
Round Robin VM allocation policy is VM allocation is based on time slices principle. This VM
allocation policy, time scheduling method enhances performance by utilizing less time to schedule
virtual machine.
8.3. Best Fit Host VM allocation policy
This VM allocation policy selection the host available according to the host's bandwidth, CPU
and memory resource information to allocate incoming VMs to balance the overall load of the data
centre .This algorithm effectively optimize the bandwidth, memory and CPU utilization of the data
centre and achieve the reduce energy consumption.
8.4. Fuzzy VM allocation Policy
This VM allocation policy is combination of best fit VM allocation policy with fuzzy logic to
allocate VMs using information about available CPU, bandwidth and memory resources and
additional feature weights to balance with the computational resource, memory and bandwidth. This
fuzzy algorithm carried out the finest optimizing the computational resource utilization of the data
centre but didn't impact on energy consumption. Our Experimental Results based on experiments
performed on all three algorithm and compares which VM allocation policy is best. According to
graphs Fuzzy based VM allocation policy are the best policies [27].

8
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

For the purpose of simulation we need to set parameter in cloudsim environment. For the purpose
for all algorithms we set the simulation parameter about host information setting, data center
information setting. After that simulation provide all the running information to the cloud simulator
and the given below information is Host parameter setting and data center parameter setting that is set
into the cloudsim simulator.

9. Simulation Parameter

These are the simulation parameters which are set for the comparative analysis carried out:
Host Information
• Ram = 36000;
• Storage = 1000000
• Bandwidth = 40000;
The Properties of a data center: architecture, OS, list of
Machines, allocation policy:
• Operating System: CentOs
• System architecture: x86
• VM: Xen
In this analysis simulation and setup for VM allocation based on fuzzy sets, advanced VM allocation
and simple VM allocation. We simulate three policies Simple, Advance, and Fuzzy allocation based.
In this work studied and examined numerous load balancing policies for the cloud environment.
These load balancing algorithms basically depend on VM allocation to the host. The simulation of
various VM allocation policy algorithms and testing of algorithms has been developed in CloudSim
environment. The simulation results shown Best Fit VM allocation policy makespan is better than
others VM allocation policy algorithm. The load balancing was completed in context of availability of
RAM and necessities of hosts and VMs, respectively. We simulate these algorithms in CloudSim
environment. These algorithms comparison based on various parameters is given in table-1. The
comparison table is showing the information about performance of algorithms. We evaluate the result
of above define algorithms based on CPU utilization, Memory utilization and bandwidth utilization.
This is simulation result that we get after the simulation has been carried out.

10. Result Analysis

Results for Simple VM allocation Policy

SD of Mean of CPU utilization 2661.0649068541006
SD of Mean of Bandwidth utilization 6036.022047093209
SD of Mean of memory utilization 1788.3977508373248
Average no of hosts 24
Results for Advanced VM Allocation Policy
SD of Mean of CPU utilization 3184.266757631668
SD of Mean of Bandwidth utilization 7151.87080037969
SD of Mean of memory utilization 2131.2980157641036
Average no of hosts 13

9
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

Results for fuzzy set based VM allocation Policy

SD of Mean of CPU utilization 2661.0649068541006
SD of Mean of Bandwidth utilization 6036.022047093209
SD of Mean of memory utilization 1788.3977508373248
Average no of hosts 24
These graphs are showing the result of VM allocation policy experimental result. The Figure 3-6 is showing
performance. The overall performance metrics are CPU utilization, memory utilization, bandwidth utilization
and average host for VM allocation policy. These results are cumulative in performance table. This comparison
table showing the result value for each algorithm. The results of algorithms are showing the information about
the average hosts for each algorithm. Figures are showing the comparison information of each algorithm.
According to simulation result Fuzzy algorithm results are better than other algorithm. The simulation of VM
allocation policies we set the Number of VM =500 and cloudlet = 1000.

TABLE I. COMPARISON TABLE FOR RESULT PARAMETER

Simple VM Advanced VM Fuzzy set based
Performance
allocation allocation VM allocation
Parameter
Policy policy Policy
Mean of CPU
2661.064907 3184.266758 2661.064907
utilization
Mean of
Bandwidth 6036.022047 7151.8708 6036.022047
utilization
Mean of
memory 1788.397751 2131.298016 1788.397751
utilization
Average no of
24 13 24
hosts

This figure 3 is showing the overall performance of all VM allocation algorithms. This figure 4 is showing the
CPU utilization, memory utilization and Bandwidth utilization. This graph is showing one additional information
i.e. Average number of host.
This figure 5 is showing the mean of CPU utilization used by each VM allocation policy algorithm after the
simulation, we calculated the standard deviation that is representing as SD(standard deviation). This chart is
showing the fuzzy VM allocation is the best among policies algorithm.

Figure 3. CloudSim Classes

10
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

Figure 4 is showing the use of CPU utilization for different VM allocation policy. This figure 5 is showing
the Bandwidth utilization used by each VM allocation policy algorithm. This chart is showing the fuzzy VM
allocation is the best among policies algorithm.
This figure 6 is showing, standard deviation of memory utilization for different VM allocation policy.
In our experimental result we calculate the standard deviation of memory, CPU and bandwidth utilization by
the VM allocation policies.

Fig.4. Graph for Bandwidth utilization Fig.5. Graph for CPU utilization

Fig. 3. CloudSim core simulation engine

Fig. 4. CloudSim core simulation engine

Fig.6. Graph for Memory utilization

This comparison table information gathered from simulated results and CPU, Bandwidth and memory
utilization in the form of standard deviation of values. These all values are showing in the table and graph for all
three VM allocation policies algorithms results. According to the graph and result, comparison table, fuzzy logic
is best policy.

11
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

11. Conclusion

In this research paper we studied and carried out the analysis of the VM allocation policies in the CloudSim
environment. We simulate the environment for the VM allocation policies and according to simulation result
fuzzy VM allocation policies is best explained in literature. In this paper we evaluate different parameter for
each VM allocation policy algorithm. According to simulation results, in this research paper VM allocation
policy if done through fuzzy based approach is yielding good results.

References
[1] S. Sonkar and M. Kharat, “A Review on Resource Allocation and VM Scheduling Techniques and a Model
for Efficient Resource Management in Cloud Computing Environment,” in ICTBIG, Indore, India,
Nov. 18-19, 2016, pp. 1-7, doi: 10.1109/ ICTBIG. 2016.7892646.
[2] M. Ullrich et al., “Towards Efficient Resource Management in Cloud Computing: A Survey,” presented at
the 2016 IEEE 4th Int. Conf. FiCloud, Vienna, Austria, Aug. 22-24, 2016, pp. 170- 177, doi: 10.1109/
FiCloud.2016.32.
[3] S. Kadioglu et al., “Heterogeneous Resource Allocation in Cloud Management,” presented at the 2016
IEEE 15th Int. Symposium on NCA, Cambridge, USA, Oct. 31-Nov. 2, 2016, pp. 35-38, doi:
10.1109/NCA.2016.7778589.
[4] S. Aslam and M. Shah, “Load Balancing Algorithms in Cloud Computing: A Survey of Modern
Techniques,” presented at the 2015 NSEC, Rawalpindi, Pakistan, Dec. 17, 2015, pp. 30-35, doi:
10.1109/NSEC.2015.7396341.
[5] A. Gupta and R. Garg, “Load Balancing Based Task Scheduling with ACO in Cloud Computing,”
presented at the 2017 ICCA, Doha, UAE, Sept. 6-7, 2017, pp. 174-179, doi:
10.1109/COMAPP.2017.8079781.
[6] . Khara and U. Thakkar, “A Novel Approach for Enhancing Selection of Load Balancing Algorithms
Dynamically in Cloud Computing,” presented at the 2017 Comptelix, Jaipur, India, Jul. 1-2, 2017, pp.
44-48, doi: 10.1109/COMPTELIX.2017.8003935.
[7] M. Rathore, S. Rai and N. Saluja, “Load Balancing of Virtual Machine Using Honey Bee Galvanizing
Algorithm in Cloud”, International Journal of Computer Science and Information Technologies, vol.
6(4), IJCSIT,2015.
[8] Abhay Kumar Agarwal, Atul Raj, “A New Static Load Balancing Algorithm in Cloud Computing”,
International Journal of Computer Applications (0975 – 8887) Volume 132 – No. 2, December 2015.
[9] Alharbi, F., Rabigh, K.S.A., 2012. Simple scheduling algorithm with load balancing for grid computing.
AsianTrans. Comput.2(2),8–15.
[10] AlShawi, I.S., Yan, L., Pan, W., Luo, B., 2012. Lifetime enhancement in wireless sensor networks using
fuzzy approach and A-star algorithm. IEEE Sens. J. 12 (10),3010–3018.
[11] Amazon EC2 Reserved Instances, http://aws.amazon.com/ec2/reserved-instances,2012.
[12] Armstrong, R., Hensgen, D., Kidd, T., 1998. ’The relative performance of various mapping algorithms is
independent of sizable variances in run-time predications’, In: 7th IEEE Heterogeneous Computing
Workshop (HCW98), Mar. 1998, pp.79–87.
[13] Azar,Y., Cohen,I.R., Fiat,A., Roytman,A., 2016,January. ’Packing small vectors’, In: Proceedings of the
Twenty-Seventh Annual ACM-SIAM Symposium on Discrete Algorithms, pp.1511–1525.
[14] Berl, A., Gelenbe, E., Di Girolamo, M., Giuliani, G., De Meer, H., Dang, M.Q., Pentikousis, K., 2010.
Energy-efficient cloud computing. Comput. J. 53 (7), 1045–1051.
[15] Bohn, R.B., Messina, J., Liu, F., Tong, J., Mao, J., 2011. NIST Cloud Computing Reference Architecture,
IEEE World Congress on Services, Washington, DC, pp. 594–596.
[16] Calheiros, R.N., Ranjan, R., Beloglazov, A., De Rose, C.A., Buyya,R.,2011.CloudSim: a toolkit for
modelling and simulation of cloud computing environments and evaluation of resource provisioning
algorithms. Soft. Pract. Exp. 41 (1), 23–50.
[17] Chaisiri, S., Lee, B.S., Niyato, D., 2012. Optimization of resource provisioning cost incloud computing.
IEEE Trans. Serv. Comput. 5 (2), 164–177.Chen, H., Wang, F., Helian, N., Akanmu, G., 2013. User-
priority guided Min-Min scheduling algorithm for load balancing in cloud computing. IEEE Natl.
Conf.Parall. Comput. Technol. (PARCOMPTECH),1–8.
[18] Chien, N.K., Son, N.H., Loc, H.D., 2016, January, ’Load balancing algorithm based on estimating finish
time of services in cloud computing’, In: 2016 18th IEEE International Conference on Advanced
Communication Technology (ICACT), pp. 228–233.

12
ICAPSM 2021 IOP Publishing
Journal of Physics: Conference Series 2070 (2021) 012129 doi:10.1088/1742-6596/2070/1/012129

[19] Dam, S., Mandal, G., Dasgupta, K., Dutta, P., 2015, February. ’Genetic algorithm and gravitational
emulation based hybrid load balancing strategy in
cloud computing’, Third IEEE International Conference on Computer, Communication, Control
and Information Technology (C3IT), pp.1–7.
[20] Dasgupta, K., Mandal, B., Dutta, P., Mandal, J.K., Dam, S., 2013. A genetic algorithm (ga) based load
balancing strategy for cloud computing. Procedia Technol. 10,340–347.
[21] Deng, Y., Lau, R.W., 2014. ’Dynamic load balancing in distributed virtual environments using heat
diffusion’, ACM Transactions on Multimedia Computing, Communications, and Applications
(TOMM), 10(2), pp. 16.
[22] Deng, X., Wu, D., Shen, J., He, J., 2016. Eco-aware online power management andload scheduling for
green cloud datacenters. IEEE Syst. J. 10 (1), 78–87.
[23] Garg, S., Dwivedi, R.K., Chauhan, H., 2015, September. ’Efficient utilization of virtual machines in cloud
computing using Synchronized Throttled Load Balancing’, 1st IEEE International Conference on Next
Generation Computing Technologies (NGCT), pp. 77–80.
[24] Glover, F., Laguna, M., 2013. Tabu Search. Springer, New York, pp. 3261–3362.
[25] Hadji, M., Zeghlache, D., 2012, June. ’Minimum cost maximum flow algorithm for dynamic resource
allocation in clouds’, IEEE 5th International Conference on Cloud Computing (CLOUD), pp. 876–
882.
[26] Hsiao,H.C., Chung,H.Y., Shen,H., Chao,Y.C., 2013. Load rebalancing for distributed file system sin
clouds. IEEE Trans. Parall. Distribut. Syst.24 (5), 951–962.
[27] Hwang, C.R., 1988. Simulated annealing: theory and applications. Acta Appl. Math.2 (1), 108–111.

13