Load Balancing in Cloud Computing: Algorithms, Types, and Analysis

This report provides a comprehensive overview of load balancing in cloud computing, a critical aspect for managing resources and ensuring optimal performance. It begins by defining cloud computing and the need for load balancing to distribute workloads efficiently across multiple nodes. The report details the goals of load balancing, including system stability, fault tolerance, and improved performance. It then categorizes load balancing types, distinguishing between static and dynamic approaches. The core of the report focuses on various load balancing algorithms, such as Round Robin, Min-Min, Opportunistic Load Balancing, Max Min, Active Monitoring, Equally Spread Current Execution, Active Clustering, and Throttled Load Balancing. Each algorithm is explained with its operational mechanics, advantages, and disadvantages. The report concludes by summarizing the importance of load balancing in cloud environments and its role in enhancing resource utilization and user satisfaction.

Running head: CLOUD COMPUTING
CLOUD COMPUTING
Name of the Student:
Name of the University:
Author Note:

1CLOUD COMPUTING
Table of Contents
Introduction......................................................................................................................................2
Load Balancing................................................................................................................................2
Goals............................................................................................................................................3
Demand........................................................................................................................................3
Types............................................................................................................................................4
Needs...........................................................................................................................................4
Algorithms.......................................................................................................................................5
Round Robin Algorithm..............................................................................................................5
Min-Min Load Balancing Algorithm...........................................................................................6
Opportunistic Load Balancing Algorithm...................................................................................6
Max Min Load Balancing Algorithm..........................................................................................7
Active Monitoring Load Balancing Algorithm...........................................................................7
Equally Spread Current Execution Algorithm.............................................................................7
Active Clustering Algorithm.......................................................................................................8
Throttled Load Balancing Algorithm..........................................................................................8
Conclusion.......................................................................................................................................9
References......................................................................................................................................10

2CLOUD COMPUTING

3CLOUD COMPUTING
Introduction
Cloud Computing grew as favorites in recent times. Due to service’s part, this provides
flexible in retrieving data and easy way for keeping files for making large files and sets of data
accessible for numerous consumers to overall world. Managing such kinds of huge sets of data
call for many approaches for enhancing and simplifying operations as well as provide perfect
efficiency levels for consumers. Load balancing is technique which distributes workload among
several nodes within presented workspace so that this ensures no nodes within system is idle or
overloaded for every moment (Ghomi, Rahmani & Qader, 2017). Efficient algorithm of load
balancing might clarify each single node within system might have less or more identical
quantity of work.
Accountability of algorithm of load balancing is this is managing assignments that are
ahead of cloud area of unused services. Hence, overall accessible time for reactions could be
enhanced. Additionally, this gives proficient use of resources. Balancing workload continue as
one of worries within cloud computing as quantity of the demands could not be figured out
which are released in cloud environment (Mesbahi, Hashemi & Rahmani, 2016). Load
balancing’s fundamental consideration within platform of cloud is in appointing as well as
distributing load dynamically through nodes with certain end goal for satisfying consumer
necessities as well as for giving optimal use of resource by arranging overall obtainable load into
diverse nodes.

4CLOUD COMPUTING
Load Balancing
It is fact to disperse load through many resources within each system. In such manner,
load must be distributed over resources in construction modeling based on cloud, as for all
resources do around task’s identical quantity at each point of time. The elementary need is
delivering few approaches for stabilizing demands for giving choice of application quicker (Joshi
& Kumari, 2016). Method of load balancing which makes every processor busy along with for
completing works around within same time.
Goals
Load balancing’s goals are as follows:
 System’s stability remains in track.
 Have capability in altering this as per extend or modification within setup of system.
 Promote system of fault tolerant with respect to stamina, performance under system’s
partial failure.
 Achieve huge improvement within performance.
 Increase system’s adoptability for adjusting to modifications.
Demand
It is technique where this delegates for accomplishing task equally in each attainable
node which is shown within the system. Fulfillment of high user is aspect definitely around this.
Due to client’s large number and their needs are growing, clouds would need in supplying
products to visitors. Desirable process of load balancing helps in utilizing available resources,
verifying not any node is under load or over load (Hota, Mohapatra & Mohanty, 2019). It
facilitates scalability, minimizes time period that is consumed for actually giving responds and

5CLOUD COMPUTING
prevents difficulties. Numerous algorithms of load balancing are developed during the order for
planning load among maximum machines.
Types
Load balancing could be classified as system’s current state, like dynamic and static load
balancing.
Static
Static load balancing is knowledge of past related to resources and software of the
system. Choice for moving workload should not exactly depend on system’s present status. It
relates with load balancing which distributes workload derived which is related to policies’
constant set, associated to qualities of workload. It is not defensive. Hence, each system has
minimum one allotted assignment for itself (Lakhani & Agrawal, 2018). The static algorithms
don’t consider dynamic modifications during runtime. Few algorithms of static load balancing
are Min-Max, Round Robin and Min-Min algorithm.
Dynamic
It doesn’t consider system’s prior state and no previous understanding is needed. It
depends on machine’s present status. Usual method is permitted by it for relocating from the
machines that are heavily loaded dynamically for obtaining quick execution. In such conditions,
there is rise of case communication and turns in more if there is enhancement in variety of the
processors (Mukati & Upadhyay, 2019). Few dynamic load balancing are the honey-bee
foraging, joint-idle queue, biased random sampling and active clustering.

6CLOUD COMPUTING
Needs
Most people could equalize system’s work through relocating workload dynamically to
system nearby to the faraway nodes or the systems which are used less. Performing this enhances
client’s fulfillment, reducing time of reaction, maximizing utilization of resource, decreasing set
of quantity of refusals of tasks and boosting system’s efficiency stability. In addition, green
computing within cloud could be achieved by using the load balancing (Islam & Hasan, 2017).
Load balancing could cut down easily capacity of utilization of power by keeping distance from
machines’ overheating due to excess workload. Carbon discharge and energy absorption are two
sections for same stage. These are both specifically relative with each other. Minimizing
utilization of energy by use of load balancing would cut down automatically discharge of carbon
and for such reason produce green computing.
Algorithms
Following algorithms of load balancing are used currently within cloud computing:
Round Robin Algorithm
It is algorithm for static load balancing which uses fashion of round robin to allocate jobs.
This scheduling is quite efficient and effective time scheduling policy. The algorithm randomly
selects nodes for load balancing. Here, essential role is played by data centers in handling
process of the load balancing within cloud computing (Shakir & Razzaque, 2017). When data
center’s controllers receives request from the user, then this passes request to algorithm of round
robin. Within the algorithm, there is division of time in small units which is known as time slice.
Hence, the algorithm is specially designed for sharing of time.

7CLOUD COMPUTING
Firstly, every processor which could be run, are stored inside circular queue. In defined
slot of time, server is allocated by scheduler to every process within queue. When there are new
processes, this would be added at queue’s end. First process is selected by scheduler from queue
randomly. As there is end of time slot of process, process is passed on from server and then
attached at queue’s tail. If this process is totally completed before time slot, process is voluntary
released by it. Server is assigned by scheduler to ready the process within queue. In such way,
there is processing of user request in a circular way through using the algorithm (Jain & Saxena,
2016). However, due to server’s random selection several times, few servers could be overloaded
that results in decrement of load balancing’s performance. For overcoming this issue, better
technique of allocation is introduced and is called weight round robin load balancing algorithm.
Min-Min Load Balancing Algorithm
This is fast or simple algorithm which gives improved performance. The algorithm
includes task set. There is not assigned tasks initially to any nodes. Hence minimum time for
completion is calculated for every available node within system. After calculation, task is chosen
which have minimum time for completion and assign to separate node. The time that is currently
available for execution is uploaded, then there is removal of task from available set of task. The
process is performed until every task would be allocated in equivalent machines (Xu, Tian &
Buyya, 2017). The algorithm works better for situations where there is more number of smaller
tasks than the larger tasks. This algorithm’s disadvantage is that this leads into starvation as this
assigning smaller tasks firstly, making large tasks wait in waiting stage.
Opportunistic Load Balancing Algorithm
The algorithm doesn’t analyze virtual machine’s current state as this is static load
balancing algorithm. This makes effort in keeping all nodes busy. The algorithm manages

8CLOUD COMPUTING
unexecuted tasks rapidly to nodes that are available within system. Every task could be randomly
assigned to node. The algorithm doesn’t give load balance good results (Mazher, et al., 2018).
Due to such reason, this doesn’t calculate current time of execution of node, hence task would
process slow with this manner.
Max Min Load Balancing Algorithm
The algorithm is same as algorithm of Min Min Load Balancing. At beginning every task
that is available is submitted to system and calculation is done for minimum time for completion
for every available task. After the calculation, a task is selected which have maximum time for
completion and the task is allocated to corresponding machine (Kumar & Kumar, 2019). This
algorithm’s performance is better when compared with Min Min algorithm as if only single large
task is there in task set, then short tasks are run parallel by Max Min algorithm with the large
task.
Active Monitoring Load Balancing Algorithm
It is algorithm of dynamic load balancing where load is allocated to virtual machine
through finding out least loaded virtual machine or idle virtual machine in list. Initially, there is
search for null virtual machine if no null virtual machine is there. Further the least loaded virtual
machine is chosen. Here index table for every requests and servers which are assigned to servers
currently is maintained with help of load balancer. When there is new request, the servers’ index
table is scanned by the data center which is least loaded or idle. The algorithm uses concept of
first come first serve to assign load to server having least index number for greater than two
servers (Babu & Samuel, 2016). By using the server id, load is allocated to server as well as
index table of server is incremented. After completing the task, the data center is forwarded the

9CLOUD COMPUTING
information as well as server’s index table is decremented. When a new request comes, index
table is rescanned with load balancer and process allocation takes place.
Equally Spread Current Execution Algorithm
It is algorithm of dynamic load balancing where effort is made by load balancer for
distributing equal quantity of load between every server which is available in data center. The
processes are assigned priority at starting of the algorithm, it then checks capacity and size for
transferring load to server that could handle the load in smaller time period (Milani &
Navimipour, 2016). At such point, there is measure of capacity of virtual machine and estimation
of load. Load is allocated as per capacity and size of the matching virtual machine.
Active Clustering Algorithm
The algorithm defines virtual machine’s clustering to balance load within cloud
computing. For the algorithm, the clustering is grouping of the objects together that have similar
kind of properties (Adhikari & Amgoth, 2018). Hence, virtual machines having similar
properties are together grouped in the cluster for handing kind of load.
Throttled Load Balancing Algorithm
The algorithm is about virtual machine. Throttled Load Balancer (TLB) maintains every
process as well as monitors work on servers. Hence, in the algorithm, best virtual machine is
found by load balancer for client request which could handle load in an effective way and quite
easily. Different virtual machines have different properties and capacity for handling different
loads. Hence, as per load, right virtual machine should be selected for load. There is maintenance
of maintenance of index table for every server and when the data center I sent request by the
client, data center’s controller forward request to throttled load balancing. To find idle server that

10CLOUD COMPUTING
is available, index table is scanned by TLB and send back server id to data center and task is
allocated to the servers (Khan & Ahmad, 2017). Index table after allocation is updated.
Whenever controller of data center gets information of task completion there is decrement again
in index table. In the algorithm, if no server is there in the idle state, request remains in queue.
Conclusion
Cloud computing is an emerging trend within IT’s era having huge requirements for
infrastructures, storage and resources. Load balancing is cloud computing’s essential aspect for
balancing load in system. Numerous users are allowed in accessing distributed, hardware,
software, virtualized and scalable resources over internet by cloud computing. Load balancing is
major issue for cloud computing. This is mechanism that distributes workload over every node
within overall cloud. This would improve resource utility and overall performance of system.
This paper discusses about load balancing and its goals, demands, types and needs. This paper
also analyses algorithms of load balancing within cloud computing. These algorithms of load
balancing ensure resources’ utilization through distributing load between several nodes within
system by use of task scheduling.
Implementation level in dynamic as well as static algorithms plays quite crucial role to
decide effectiveness and efficiency of algorithms. In the paper, load balancing, comparisons of
dynamic and static load balancing, demand along with load balancing’s goals are analyzed. This
work might give better guide for crucial issues which should be addresses through design of
algorithm of the load balancing within cloud computing. For future, algorithms of multiple load
balancing could be combined that would maintain better tradeoff among several criteria for
performance.

11CLOUD COMPUTING