An Optimized-Throttled Algorithm for Distributing Load in Cloud Computing

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 12 | Dec 2023 www.irjet.net p-ISSN: 2395-0072
© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 1
An Optimized-Throttled Algorithm for Distributing Load in Cloud
Computing
Tanjina Jahan1, Prof. Dr. Mohammod Abul Kashem 2, Md. Toufecul Islam 3
1Tanjina Jahan: Student, Dept. of Computer Science and Engineering, Dhaka University of Engineering and
Technology, Bangladesh
2Dr. Mohammod Abul Kashem: Professor, Dept. of Computer Science and Engineering, Dhaka University of
Engineering and Technology, Bangladesh
3Md.Toufecul Islam: Student, Dept. of Computer Science and Engineering, Dhaka University of Engineering and
Technology, Bangladesh
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Cloud computing is a new area with many
promising applications for both established businesses and
individual users. This is a development of the idea of
decentralized computing. Based on the concept of "on-
demand" services, it delivers data, applications, and
infrastructure in response to users' immediate requests [1].
Load balancing, in which the load is distributed among many
cloud servers or nodes, improves efficiency. It's the singlemost
crucial element in maximizing the use of available resources.
Load balancing has emerged as a critical process in cloud
computing infrastructures [2]. To meet the needs of such a
large user base, a distributed solution is necessary, since
centrally managing one or more idle services is neither
practical nor cost-effective. It'simpossibletogivecertainusers
control over individual computers. The "cloud" in "cloud
computing" refers to the large network of distributed nodes
[3]. As a result, it needs load balancing to distribute the
workload across its many servers or virtualmachines.Inorder
to improve cloud performance, the research suggested an
algorithm that prioritizesloaddistributiontominimizevirtual
machine overload and underload. Cloud Analyst is used to do
in-depth research and comparisons. We compared our results
to those of the more seasoned Round Robin and Throttled
algorithms. In addition, simulation findings show that the
proposed algorithm has improved response times and
processing times in the cloud data center, demonstrating its
superiority over current methods.
Key Words: Cloud Computing, Load Balancing, Processing
Time, Response Time.
1. INTRODUCTION
Corporations and educational institutions use
distributed computing's new cloud computing
technology to store and access data. The key difficulty
is planning for incoming queries in a way that
minimizes response time while optimizing resource
use. In order to process client requests with the
quickest feasible response time and allocate them to
virtual machines, a number of algorithms are utilized,
including FCFS, Round Robin, and Throttled. Many
resources are not involved intheexecutionofrequests,
and the cloud system is unbalanced because of
restrictions such as excessive communication delays
and underutilization of resources [4].
Load balancing is essential for maximizing throughput
and reaction time in a cloud environmentbecauseeach
virtual machine performs the same function.Toevenly
distribute work between computers, we may
dynamically move tasks from one machine to another.
This enhances the system's performance ratio,
optimizes user satisfaction, decreases response time,
maximizes resource usage, and decreases task
rejections. Virtualization technology can effectively
manage the dynamic resources of a cloud-based
platform. By allowing severalvirtualmachinestoshare
a single physical server, it introduces a novel approach
to enhancing the power efficiency of datacenter
consolidation. This means that the cloud computing
system's energy usage may be lowered by turning off
or putting some of the servers into sleep mode [5].
In this research, we provide a new approach for
distributing incoming jobs across available virtual
machines. Here, new requests are sent to the least-
loaded VM once the least-used VM has completed its
current workload. This technique is compared to the
more traditionalround-robinandthrottledapproaches
to reducing virtual machine underutilization. The
proposed algorithm focuses on efficiently dividing up
incoming work across available virtual machines.
When there are none in the VM index, it looks for the
least busy ones and adds them.

2. ROUND ROBIN ALGORITHM
One of the simplest static load-balancing algorithms is
the round-robin algorithm, which simply rotates
amongst workers in a loopuntil all requests have been
processed. The request is then sent to a pool of
randomly chosen virtual machines by the data center
controller. All servers and nodes are grouped together
into clusters according to how quickly they handle
data. When a virtual machine (VM) is selected for a
task, it moves to the front of the queue. However, as
long as the virtual machine isn't empty, new incoming
workloads must wait in line, which is the core problem
with this allocation. Because of this issue, resource
management would suffer, leading to longer response
times and decreased productivity. The Round Robin
algorithm prevents feeling full almost entirely while
having low throughput [6].
Fig -1: Round Robin Algorithm
3. THROTTLED LOAD BALANCING ALGORITHM
(TLB)
In this approach, the load balancer is in charge of
updating the index table. The virtual machine's
availability (busy or not) is shown. When a job comes
in, a load balancer sends it to the right virtual machine.
Which may be used to fulfil the request made by the
user. However, when throttled checks for available
virtual machines, it starts at index in the database [6].
Fig -2: Throttled Algorithm
4. OPTIMIZED THROTTLED ALGORITHM (OTA)
Step 1: In the first phase of the load balancing
procedure, the load balancer maintains and refreshes
an index table that contains the availability (free, "0")
or occupation (occupied, "1") of all VMs. Virtual
machine IDs start at '0'.
Step 2: Data Center Controller received a request and
then requested the new assignment from the Load
Balancer.
Step 3: The load balancer checks the index table and
determines that the first available VM is free to be
used.
The following actions are taken if a VM is located:
▪ The VM's ID is sent to the VM index.
▪ The data center controller receives the VM ID
from the VM index.
▪ When a fresh allocation becomes available, the
data center controller notifies the load balancer.
▪ After receiving requests from the data center
controller, the load balancer updates the index
and waits for them to complete.

If a virtual machine (VM) cannot be located:
▪ Identifies the least busy virtual machines and
adds them to the VM index.
▪ The ID of the least-loaded virtualmachineissent
to the data center's control system through the
formula. VMloadi < = CVMi ∗ TLL
Where CVM=Penum*Pemips and VMloadi=
▪ VM Index transmits the VM ID to the DataCenter
Controller.
Step 4: The controller goes back to Step 2 and
continues working there.
Fig -3: Optimized Throttled Algorithm
5. RELATED WORK
The following section offers a high-level summary of
the load balancing approaches made use of in cloud
computing. The main goal is to distribute all incoming
tasks to available virtual machines so they may be
processed immediately.
To ensure maximum utilization of each server-created
virtual machine on cloud platforms, Garg et al. [7]
introduced the Synchronized Throttled VM Load
Balancing (STVMLB) Algorithm, which focuses on load
balancing to reduce the possibility of overload and
underload on virtual machines. This is a more effective
approach to synchronizing all virtual computers. By
adapting the basic idea behind Throttled, a load
balancing algorithm, the system was created, and it
increases virtual machine utilization even higher than
Throttled and Active Monitoring. The suggested
dynamic load-balancing technique performs well in a
cloud setting, evenly distributing requests and
maximizing responsetime.TheThrottledalgorithm,on
the other hand, reduces the speed of response.
In this research, N. Xuan Phi et al. [8] evaluate an
existing method and propose an enhanced versionofit
to better balance loads than previous algorithms. By
recommending this strategy, we were able to
accomplish our goals of decreasing the backlog of
inquiries waiting to be sent and speeding up the
processing and response time of cloud data centres in
contrast to two previous methods. It also shows that
the suggested method is more efficient than Round
Robin and Throttledwhenitcomestocloudcomputing.
The goal of this study is to improve end-user
performance in cloud computing by proposing a
throttled modified algorithm (TMA) for increasing the
response time of virtual machines.
To avoid eitheroverburdeningorunderutilizingvirtual
machines, G. Soni et al. [9] offer a Central Load
Balancer (CLB) technique. Using criteria such as
priority and status, CLB divides work among many
virtual computers. Author simulations showed that
CLB-based load balancingalgorithmsperformedbetter
than Round Robin (RR) and Throttled algorithms.
Although doing so would provide for a steadier and
more dynamic load distribution, thesuggestedmethod
does not factor in the present utilization of resources
like CPU and memory.
In this study, A.A. Alkhatib et al. [10] present a
thorough examination of load balancingstrategiesthat
use a wide variety of VMs to accomplish the task.
Several static and dynamic load balancing techniques,
along with their benefits anddrawbacks,arepresented
in this work. Furthermore, this study offers a
comparison of several load-balancing techniques.
There is a unique identification and a summary of the
major qualities and limits of each load balancing
method. The purpose of this study is to investigate the
notion of load balancing in cloud computing as well as
the advantages and disadvantages of a specific load
balancing method. Algorithms including "honeybee
foraging," "throttled," "Min-Min," "weight,""Max-Min,"
and "round robin" are among those studied, as are the
metrics and issues related to them.

With the aid of a distinctive VM-assign load balancing
technique, S. G. Domanal et al.'s [11] study evenly
distributes new requests among all available virtual
machines. In this situation, the virtual machine is
allocated depending on how it will be used. The first
request is given to the virtual machine that has
received the fewest requests. This technique is then
compared to the currently used Active-VM algorithm,
which has been shown to significantly increase virtual
machine underutilization. The results are analyzed to
prevent under- and over-loading of virtual devicesand
guarantee their full usage. We also comparetheresults
to the current Active-VM load balancing techniqueand
assess its effectiveness using the CloudSim simulator.
However, it cannot respond in real time to changes in
user input.
The LDAB scheduling method, used for QoS and load
balancing, was created by Atyaf Dhari et al. [12]. Load
balancing is essential for keeping the whole system
stable. The system's efficiency may be improved,
therefore, bydistributingthetaskacrossseveralvirtual
machines. The goal of the suggested load balancing
algorithm (LBDA) is to decrease processing and
response times while maintaining workload parity
across a data centre's virtualmachines.Therearethree
phases to the LBDA's operation: First, determine the
VM's capabilities and workload to determine the VM's
current stage. Figure out how long it will take to do the
task using all of the virtual machines.Dependingonthe
availability of virtual machines and the timing of tasks,
decide how to distribute the workload among them.
Three other algorithms—MaxMin, Shortest Job First,
and Round Robin were used to evaluate this one. In
comparison to thesemethods,LBDAproducessuperior
results.
Cloud computing and load balancing are two of the
cloud computing resource allocation approaches
identified in the current research by S. H. Sabeti et al.
[13]. The author places an emphasis on load balancing
and tries to ensure that all servershaveaboutthesame
amount of work to do. To speed up responses and
processes, this research suggests using a load-
balancing algorithm that combines elements of the
ESCE and Throttled algorithms. To reduce the time
spent checking for a suitable virtual machine that can
handle longer tasks and improve response time, the
algorithm first proposes the least busy machine. Two
more virtual algorithms, Throttled and ESCE, are
combined into a single hybrid algorithm that is
proposed. All four algorithmsweresimulatedusingthe
same framework, and the results showed that the
suggested method completed tasks more quickly and
had a lower total number of iterations than the other
three. Additional goals, such as reduced costs and
enhanced performance, have not yetbeenattaineddue
to scheduling and technical constraints.
The Cloud Analyst Simulator was used to evaluate the
Modified ThrottledLoadBalancingAlgorithm,theFCFS
Algorithm, and the Particle Swarm Optimization
Algorithm by P. A. Pattanaik et al.[14].Accordingtothe
findings, Particle Swam is the optimalization method
that yields the quickest response time comparedtothe
other two. Moreover, Particle Swam optimization has
lower total server costs than the other two techniques.
Since costs play a major role in the cloud, minimizing
them should be a key concern in terms of both
efficiency and customer happiness. Using the Particle
Swam Optimization Algorithm, we were able to find a
better distribution map that represents the ideal
option for our resources. The simulation outcomes are
recorded in terms of response time, datacenter
processing time, efficiency, and arrival costs for all
three methods.
The Modified Throttled method, introduced by S. G.
Domanal et al. [15], is similar to the Throttled
algorithm in that it maintainsanindextableofVMsand
VM states. Response times have been improved, and
free virtual machines have been used to their full
potential. The proposed technique is a way to choose a
VM to handle a client's request, with the VM at the first
number being picked mostly based on its state. If the
virtual machine is available, its ID is transmitted to the
data center with the query;otherwise,-1isreturned.In
contrast to the Throttled algorithm, which parses the
index table about the first index each time the data
centre asks the load balancer to assign VMs, the next
VM at the index next to the currently allocated VM is
chosen based on its status when the next query comes
in.
In this work, S. Y. Mohamed et al. [16] present the
Balanced Throttled Load Balancing (BTLB) method.
Results fromotherloadbalancingalgorithms,including
round robin and AMLB, as well as the throttled load
balancing algorithm, are compared with those from
BTLB. All four of these algorithms' efficacy will be
shown in this analysis. The proposed technique is
shown to decrease response times. The results were
calculated using a cloud analyst simulation. After
comparing simulation results with the four methods,
the author may conclude that the Balanced Throttled

load balancing approach has the fastest average
response time.
6. SIMULATION AND EVALUATION
New algorithms and approaches to cloud computing
must be thoroughly verified in a simulation before
being deployed in the real world. It may be quite costly
and time-consuming to create a functional cloud
computing environment in which to evaluate freshly
suggested algorithms and methods. As a result, the
simulator is used to model both the cloudenvironment
and the implementation of policies.Thesimulatordoes
a great job of finding algorithms with little time and
money spent on it [17]. In this research, we used the
Cloud Analyst simulation tools together with the
Round-Robin and Throttled algorithms to model and
assess the suggested (OTA) method. We assess
variables such as the total response time of the cloud
system and the processing time of the data center.
A cloud analyst's main functions are as follows
[17]:
▪ The user-friendly interface of Cloud Analyst
facilitates experimentation.
▪ It's easy to run several tests with the same or
different settingsandseetheoutcomesvisually
using Cloud Analyst.
▪ A cloud analyst has remarkable flexibility and
customization options.
Fig -4: Snapshot of Simulator
6.1 Simulation Setup
Table -1: UserBase Settings Variable
The best way to check an algorithm is via simulation.
Where:
Peak Hour: Period of day with the most users online
Average Peak Users: The typical number of people
using the service during peak hours.
Average Off-Peak Users: Number of users on average
who log in outside of peak hours.
7. RESULTS AND ANALYSIS
Here, run the simulation three times, representing
two distinct strategies. Particularly:
▪ When using the simulator for the firsttime,itis
recommended that the Round Robin policy be
used.
▪ To use the Throttled policy for a second time.
▪ The third time around, implement the policy
using our suggested OTA methodology.
7.1 Case 1: Run a simulation using 25 simulated
virtual machines.
For case 1, we picked six user bases that correspond
to the six geographical areas of the world and five data
centers. Many users visit the application after workfor
around two hours each evening. Every internet user
makes a new request every five minutes. Each data
center includes five virtual computers, and each user
base has the characteristics indicated in Table -1.

Table -2: Parameters for Virtual Machines
Table -3: Specifics about the Inner Workings of Each Data
Center
Fig -7: Optimized Throttled Algorithm (OTA)
Chart -1: The End Outcome of a Simulation Using 25
Virtual Machines
Our OTA method is faster at both processing at the
Data Center and responding to user input than the
Throttled and Round-Robin algorithm, but only by a
small margin. That's why we ran another round of
comparisons using 50 virtual PCs with the identical
settings as before.
7.2 Case 2: Run a simulation using 50 simulated
virtual machines.
For case 2, we picked six user bases that correspond
to the six geographical areas of the world and five data
centers. A large number of users visit the application
after work for around two hours each evening. Every
internet user makes a new request every five minutes.
Each data center includes 10 virtual computers, and
each user base has the characteristics indicated in
Table -1.
Table -4: Parameters for Virtual Machines
Table -5: Specifics about the Inner Workings of Each Data
Center

Fig -10: Optimized Throttled Algorithm (OTA)
Chart -2: The End Outcome of a Simulation Using 50
Virtual Machines
Based on the experiments' outcomes in the first two
cases, the results show that OTA is faster than both of
the other methods when it comes to data center
processing and system responsiveness.Loadbalancing
is improved by the OTA algorithm compared to the
Throttled and Round-robin methods.
8. CONCLUSIONS
When it comes to improving corporate efficiency and
customer happiness withcloudcomputing,minimizing
response times is the biggest issue. Throttled Load
Balancing Algorithm, Round Robin Algorithm, and
Proposed Algorithm are three of the most well-known
dynamic load-balancing algorithms; we compared
them while keeping these considerations in mind. The
proposed algorithm's response time was found to be
faster than that of the other two techniques. The OTA
algorithm was created by expanding on the ideas
behind the Throttled algorithm for balancing load. In
order to find the best method for allocating our
resources, which will lead to a more efficient
distribution map, we have taken into account the
Proposed Algorithm. New modified improvisation
algorithms may be developed and implemented in the
real world as the focus of future studies.
REFERENCES
[1] A. M, N. N. Sharma, and M. A. S., “An Enhancement of
Throttled Load Balancing Algorithm in Cloud using
Throughput,” International Journal ofCircuitTheoryand
Applications, vol. 9(15), pp. 7603–7611, 2016.
[2] G. J. Mirobi and L. Arockiam, “DynamicLoadBalancing
Approach for Minimizing the Response Time Using An
Enhanced Throttled Load Balancer in Cloud
Computing,” IEEE Xplore, Nov. 01, 2019.
[3] A. Jyoti, M. Shrimali, S. Tiwari, and H. P. Singh, “Cloud
computing using load balancing and service broker
policy for IT service: a taxonomy and survey,” Journal of
Ambient Intelligence and Humanized Computing, vol. 11,
no. 11, pp. 4785–4814, Feb. 2020.
[4] E. Jafarnejad Ghomi, A. Masoud Rahmani,andN.Nasih
Qader, “Load-balancing algorithms in cloud computing:
A survey,” Journal of Network and Computer
Applications, vol. 88, pp. 50–71, Jun. 2017.
[5] H. Ren, Y. Lan, and C. Yin, “The load balancing
algorithm in cloud computing environment,” Dec. 2012.
[6] Amrutanshu Panigrahi,B. Sahu,SarojKumarRout,and
Amiya Kumar Rath, “M-Throttled: Dynamic Load
Balancing Algorithm for Cloud Computing,” pp. 3–10,
Oct. 2020.
[7] S. Garg, R. K. Dwivedi, and H. Chauhan, “Efficient
utilization of virtual machines in cloud computing using
Synchronized Throttled Load Balancing,” in 2015 1st
International Conference on Next Generation Computing
Technologies (NGCT), Sep. 2015.
[8] N. Xuan Phi, C. T. Tin, L. N. Ky Thu, and T. C. Hung,
“Proposed Load Balancing Algorithm to Reduce
Response Time and Processing Time on Cloud
Computing,” in International journal of Computer
Networks & Communications, vol. 10, no. 3, pp. 87–98,
May 2018.
[9] G. Soni and M. Kalra, “A novel approach for load
balancing in cloud data center,” in 2014 IEEE
InternationalAdvanceComputingConference(IACC),Feb.
2014.
[10] A. A. Alkhatib, A. Alsabbagh, R. Maraqa, and S. Alzubi,
“Load Balancing Techniques in Cloud Computing:
Extensive Review,” Advances in Science, Technology and
Engineering Systems Journal, vol. 6, no. 2, pp. 860–870,
Apr. 2021.
[11] S. G. Domanal and G. R. M. Reddy, “Optimal load
balancing in cloud computing by efficient utilization of
virtual machines,”in 2014SixthInternationalConference
on Communication Systems and Networks (COMSNETS),
Jan. 2014.
[12] A. Dhari and K. I. Arif, “An Efficient Load Balancing
Scheme for Cloud Computing,” in Indian Journal of
Science and Technology, vol. 10, no. 11, pp. 1–8, Mar.
2017.
[13] S. H. Sabeti and M. Mollabgher, “Proposing a load
balancing algorithm with an integrative approach to
reduce response time and service process time in data

centers,” in Brazilian Journal of Operations & Production
Management, vol. 16, no. 4, pp. 627–637, Nov. 2019.
[14] P. A. Pattanaik, S. Roy, and P. K. Pattnaik, “Performance
study of some dynamic load balancing algorithms in
cloud computing environment,” in 2015 2nd
International Conference on Signal Processing and
Integrated Networks (SPIN), Feb. 2015.
[15] S. G. Domanal and G. R. M. Reddy, “Load Balancing in
Cloud Computing using Modified Throttled
Algorithm,” in 2013 IEEE International Conference on
Cloud Computing in Emerging Markets (CCEM), Oct.
2013.
[16] S. Y. Mohamed, M. H. N. Taha, H. N. Elmahdy, and H.
Harb, “A Proposed LoadBalancingAlgorithmOverCloud
Computing (Balanced Throttled),” in International
Journal of Recent Technology and Engineering (IJRTE),
vol. 10, no. 2, pp. 28–33, Jul. 2021.
[17] Hamid Shoja, Hossein Nahid, and R. Azizi, “A
comparative survey on load balancing algorithms in
cloud computing,” Jul. 2014.
BIOGRAPHIES
Tanjina Jahan is pursuing her
M.Sc degree in Computer Science
and Engineering (CSE) at Dhaka
University of Engineering and
Technology (DUET).
Prof. Dr. Mohammod Abul
Kashem received his B.Sc. and
M.Sc.Engg. Degrees from State
University “Lvivska Polytechnica,”
Ukraine, in 1996 and 1997,
respectively. In 2001 he earned
Ph.D. in Control Systems and
ProcessesfromNationalUniversity
“Lviv Polytechnic” Ukraine.
Subsequently, Dr. Kashem
completed his Post Doctorate
fellowship at University Lumiera
Lyon2, France. (Erasmus Mundas
Scholarship, European
Commission) in 2016 and he was
appointed as a professorattheCSE
Department of Dhaka Universityof
Engineering and Technology
(DUET) in the year 2013.
Md. Toufecul Islam is pursuing
his M.Sc degree in Computer
Science and Engineering (CSE) at
Dhaka University of Engineering
and Technology (DUET).

An Optimized-Throttled Algorithm for Distributing Load in Cloud Computing

More Related Content

Similar to An Optimized-Throttled Algorithm for Distributing Load in Cloud Computing (20)

More from IRJET Journal (20)

Recently uploaded (20)

An Optimized-Throttled Algorithm for Distributing Load in Cloud Computing