Anda di halaman 1dari 11

IPASJ International Journal of Computer Science (IIJCS)

Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm


A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

ABC Optimized Weighted Task Load Balancing


Algorithm in Cloud Computing
Rajpreet Kaur1 and Vikas Khullar2
1
Dept. Of Computer Science and Engineering,
CT Institute of Engineering Management and Technology,
Jalandhar (India)
rajpreet.kaur503@gmail.com
2
Dept. Of Computer Science and Engineering,
CT Institute of Engineering Management and Technology,
Jalandhar (India)
vikas.khullar@gmail.com

ABSTRACT
Cloud computing is a new paradigm where high quality and low-cost information services are served by cloud services
providers on pay-per-use service. It offers scalability, availability and different services as important benefits. Cloud provides
the services to the organizations like storage, applications, and servers. The unique concept of cloud computing creates new
opportunities for Business and IT enterprises to accomplish their objectives. In cloud computing, usually, there is a number of
jobs that need to be executed with the available resources to achieve best performance, least probable total time for task
completion, a lesser processing cost, shortened average waiting time, lessened response time, and effective utilization of
resources etc. Normally errands are scheduled by client prerequisites. Latest scheduling systems should be designed to conquer
the issues proposed by the system properties in the hub of client and assets. The new scheduling methodologies might utilize a
segment of the usual scheduling recommendations to union them in conjunction with some procedure conscious techniques to
present options for higher and extra remarkable venture scheduling. Task scheduling algorithms admit being actually the
biggest hypothetical issues in the cloud computing domain. Countless deep investigations and efforts had been applied in this
regard. This paper proposes a weighted load balancing based task scheduling algorithm that considers a wide variety of
attributes in the cloud environment and uses weight based sorting for prioritizing the tasks and bee colony optimization for
balancing the load. The proposed algorithm considers four parameters i.e. Total processing cost, total processing time,
makespan time and the average waiting time. The paper desires to enhance the performance and compare the performance
with antecedent implemented task scheduling algorithm.

Keywords: Cloud Computing, Job Scheduling, Resource allocation, Efficiency, Performance, Cost, Quality of Service (QoS),
Virtual Machine (VM)

1. INTRODUCTION
A rising strategy nowadays is cloud computing. As of late, it is discovered that investigators have interest in utilizing
the cloud for carrying out technical applications and also the enormous associations are on the edge of changing over to
a hybrid cloud. A large number of applications which are very difficult need parallel processing for executing the tasks
expeditiously [1]. Because of the synchronization and communication among processes which run parallel, there is a
lessening in the utilization of resources of CPU. It is fundamental for a data center to accomplish the use of hubs while
keeping up the level of responsiveness of jobs which are running parallel. The cloud computing is pulling in an
expanded number of uses to keep running in the data centers which are remote. Numerous intricate applications
necessitate capabilities of parallel processing. A portion of the applications which are running parallel demonstrate a
decline in usage of resources of CPU at whatever point there is an expansion in parallelism only when there is no
planning of jobs accurately then it lessens the execution of a computer. [1]
According to R.Buyya that characterizes the cloud as "Cloud is a distributed and parallel system of computing which
fundamentally comprises of a collection of computers that are virtualized and inter-related and are provisioned

Volume 5, Issue 9, September 2017 Page 80


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

dynamically and presented as one or more than one unified resources of computing dependent upon service-level
agreement (SLA) set up through negotiation between the cloud's service providers and also users. Cloud computing is a
distributed model at large scale for computing, which relies on the monetary size of the operator of the cloud which is
dynamic, virtualized and abstract. The primary content of cloud computing is to oversee storage, computing power,
different sorts of services and platforms which assigned to the external clients through the internet on their interest.
Cloud computing is a quickly developing paradigm of computation with the objective of freeing up cloud's clients from
the administration of data resources, networks, software and hardware and shifting these burdens to service providers of
the cloud. [2].
1.1 Cloud Architecture
The Cloud Computing design contains numerous cloud parts; each of them is inexactly coupled. The cloud design can
comprehensively partition into two parts: Front End alludes to the customer piece of cloud computing framework. It
comprises of interfaces and applications that are required to get to the cloud computing stages, e.g., Web Browser,
Mobile, PC, and PDA. Secondly, Back End alludes to the cloud itself. It comprises of the considerable number of assets
required to give cloud computing administrations. It includes gigantic information stockpiling, virtual machines,
security system, administrations, sending models, servers, and so on. Service supplier gives the administration to the
client. These administrations are Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a
Service (SaaS) [3]. The Backend of distributed computing is only the cloud itself, which predominantly contains PCs,
stockpiling gadgets. Cloud Environment essentially comprises of programming applications that are accessed via the
web as administrations when wishes to utilize them. Applications that are based on Cloud Architectures are such that
the basic base of registering is utilized just when it is really required, draw the fundamental assets when somebody
makes an interest for those, perform a specific occupation, then give up the unneeded assets and frequently arrange
them after the employment is finished. Amid their operation, the applications scale up or down flexibly taking into
account need for assets. [4]

Figure 1 Layered View of Cloud Architecture [2]

2. SCHEDULING IN CLOUD COMPUTING


As a huge number of clients share cloud resources and dispatch their tasks to the cloud, it has become a challenge to
schedule these tasks. Hence, job scheduling is a hot topic in distributed and cloud computing. Figure 2 shows the
scheduling stages in a virtualized environment of cloud computing.

Volume 5, Issue 9, September 2017 Page 81


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

The scheduling algorithms belonging to distributed systems Job Scheduling is utilized to assign definite jobs to specific
assets specifically time. The environment of cloud computing, issue of scheduling of job is a greatest and testing issue.
Subsequently, the job scheduler ought to be rapid. The scheduling of job in cloud computing principally centers to
enhance the effective utilization of resource like lessening in completion time, memory and bandwidth. A productive
strategy of job scheduling must intend for yielding less time of response in such a way that the execution of jobs
submitted happens inside a conceivable least time and hence there would be a happening of an event of in-time where
reallocation of jobs is done. Subsequently, few dismissals of jobs happen and an additional quantity of jobs could be put
forward to the cloud by the customers which at last show expanding results in quickening the business execution of the
cloud. [1] [5]

Job scheduling is the procedure of assigning resources (or machines, such as drills, milling machines, lathes etc.) R=
{R1, R2 Rm} to jobs (or activities) A= {A1, A2 An} over sometime for fulfilling the foreordained goal. For
the most part, the allocation may conceivably be compelled by priority imperatives, asset requirements, and issue
particular limitations. The most fundamental requirements are priority imperatives, which demonstrate the job or
activity has a predefined preparing request through the resources Oi {Oi=oi1, oi2, , oin}, the established resource
limitations is the restricted limit of a machine or human asset, and confinements like non-seizure, the hinder of
procedure of production are all the issue particular limitations, which are nearly connected with the genuine condition.
For the viable environment, perhaps is uncertain, dynamic or complex, the issues of job scheduling are exceptionally
different.

Figure 2 Scheduling stages in cloud computing [13]

Cloud scheduling is categorized into- cloud service scheduling and user-level scheduling. Cloud service scheduling is
ordered at the client level and framework level. At the client level, scheduling manages issues raised by administration
procurement between suppliers furthermore, clients. The framework level scheduling handles asset administration
inside data centers. Datacentre comprises of numerous physical machines. Millions of errands from clients are
received; the task of these undertakings to the physical machine is done at the data center. This task or programming
altogether has effects on the execution sequence of datacentres. User level programming incorporates Market-based and
closeout primarily based schedulers that area unit appropriate for steering the availability and request of cloud assets.
Market-based asset distribution is successful in cloud figuring environment where assets are virtualized furthermore,
conveyed to the client as an administrator.

Volume 5, Issue 9, September 2017 Page 82


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

3. RELATED WORK
Karthick et al [6] had proposed an effective multi-queue job scheduling for cloud computing. An algorithm of MQS
(Multi-Queue Scheduling) proposed for diminishing the expenditure of both plans i.e. on-demand and reservation
utilizing the scheduler. The proposed MQS strategy gives more significance to choose job powerfully keeping in mind
the end goal to accomplish the ideal problem of cloud scheduling and thus it uses the unused free space in a financial
way. Zhang et al [7] had talked about task scheduling of cloud computing dependent upon the algorithm of Improved
CHC. In recent years, research on task scheduling in an environment of cloud computing has turned into an interesting
issue. The Improved algorithm of CHC takes far-reaching thought of the allotment of resources to tasks and the time of
completion of tasks and it likewise upgrades the speed of convergence of the algorithm. The results of experiment
demonstrate that Improved CHC algorithm has great convergence and proficiency, the average and time of completion
of the task scheduling is moderately shorter. Dave et al [8] had presented a review of various algorithms for job
scheduling in cloud computing. The algorithms of scheduling are executed by taking into account different parameters,
for example, resource availability, bandwidth, performance, physical distances, computational time, priority, cost,
latency, resource utilization, and throughput. Despite the fact that there are diverse algorithms of scheduling accessible
in cloud computing, a less near study has been done on an execution of different algorithms of scheduling regarding
aforementioned parameters. This paper goes for a near investigation of different sorts of algorithms of job scheduling
that give productive cloud administrations.

Ananth [9] had presented the survey on approaches which are related to game theory for scheduling of job in cloud
computing environment. In this paper, main concentration is on methodologies based on game theory for scheduling of
jobs in the cloud. Study of existing methodologies and different issues in the scheduling of jobs which is based on the
theory of the game is the principal goal of this paper. The methodologies are already available for scheduling of jobs
centering methodologies of game theory and investigate the open issues for examination around there. Li [10] had
solved the problems exists in multi-task scheduling in cloud computing. He proposed an improved GEP algorithm with
double fitness functions (DF-GEP) and also builds new ETCC matrixes which not only reckon the running time of all
jobs but also takes the running cost of the tasks into context. This improved algorithm reduces the optimization time
and falls into the local optimal solution hardly at the equivalent instant. This improved algorithm expresses an
excellent convergence, through experiments compared with GA and ordinary GEP algorithm by using the Map/Reduce
programming model Soni et al [11] had proposed an algorithm that firstly it stores request from the user on the basis of
Priority, Size and type of resource required. Then they, Apply Bee Colony Algorithm and assign resources. After that,
the performance is measured on the basis of requested load and served load, requested priority and served priority and
minimizing the execution time. In this, the proposed technique fulfills the user-specific requirements such as priority
execution and resource allocation.

Bansal [12] had proposed an algorithm to improve the issue of job scheduling in the environment of cloud computing.
Increasing expansion of cloud computing has enabled service providers to effectively present their services in cloud
platform; however, they still must face a prominent issue which is providing favorable quality of service parameters.
Exploiting particle swarm optimization, the proposed process shortens make-span in summation to diminishing used
memory. The results of simulation revealed that the proposed methodology considerably diminishes make-span and
memory usage in comparison to its counterparts. Gupta [13] had presented a theoretical comparison of algorithms for
job scheduling in the environment of cloud computing. This paper talks about distributed computing and its constructs
in an area (i). In segment (ii) idea of job scheduling in cloud computing has been expounded. In the segment (iii)
existing algorithms for job scheduling are talked about and are contrasted in a tabulated form and regard to different
parameters and finally, segment (iv) closes the paper giving the brief synopsis of the work. Zhang et al [14]
concentrated on the problem of job security scheduling problem under the cloud computing environment. The
framework of the cloud computing platform is made up of four sections, encompassing 1) SOA architecture, 2)
Management Middleware, 3) Resource virtualization and 4) Physical Resources. Next, formally describe the problem of
the job security scheduling under cloud computing. Finally, the experimental outcome shows that our algorithm can
successfully schedule jobs with higher level security and lesser processing time cost.

Himani [15] presented a cloud-deadline based task scheduling in cloud computing. Cloud computing is a promising
technology which generally broadens among researchers. In view of the idea of an approach to space-shared
scheduling, this work presents (CDB) Cost deadline Based algorithm of scheduling of tasks to calendar assignments
with CloudSim by considering a few parameters including user loss, provider profit, task penalty, and throughput and

Volume 5, Issue 9, September 2017 Page 83


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

task profit. Mandal [16] had proposed a task scheduling technique for cloud computing. The new strategy of task
scheduling is proposed in view of the aggregate order for the allocation of a resource to enhance the algorithm of Min-
Min. The main concentration is on minimizing the aggregate executing time (makespan) of scheduling of task and
expanding the utilization of assets. The results of the experiment show that the proposed approach allows more
versatile assets distribution for free occupations booking in the distributed computing environment. Lakra [17] had
proposed Multi-objective Task Scheduling algorithm to map the tasks to VM's which improves the throughput of the
datacenter. Tasks are sorting using Non-Dominating Sort and Machines are sorting according to the MIPS. Two task
scheduling algorithms FCFS, Priority scheduling is compared with the proposed technique. This paper concludes that
the proposed algorithm performs better than the other two algorithms having minimum execution time and increased
throughput.

Thomas [18] had proposed an improved scheduling algorithm after analyzing the traditional algorithms which are
based on user priority and task length. Tasks having high priority are not given any special importance when they
reach. The experimental research shows a significant enhancement in the utilization of resources efficiently in cloud
computing. Raja [19] proposed an improved credit-based scheduling algorithm using the parameters viz. priority, task
length, and the deadline had been proposed. In the experimental result, the proposed algorithm shows the better result
than the existing algorithm. Credits are used to reduce the makespan of the task and execute the entire tasks in the
cloud. Zalavadiya [20] proposed technique shows the priority tasks are removed from the overloaded virtual machine
and they are allocated to under loaded virtual machine. It helps to reduce the minimum completion time, amount of
waiting time of the task in a queue is minimal and achieve better resource utilization. This paper concludes that the
minimum amount of time is taken to execute the tasks and better resource utilization.

4. METHODOLOGY AND PROPOSED WORK


The existing Multi-objective Task Scheduling Algorithm [17] is used to sort the task on the basis of priority and size
but the problem arises when two tasks cannot dominate each other. Here at the place of one task, both tasks got added
to the execution list. At the same time, the same priority and same sized task occurred, then the question arises which
task has to execute first. This task selection dilemma related to weighted task scheduling algorithm [19], Multi-
objective Task scheduling algorithm has been used in our proposed algorithm. Other gap had identified in weighted
task scheduling that, the weights had been assigned to the task on the basis of three parameters such as Length,
Priority, and Deadline. The Problem found in weighted task scheduling algorithm [19], that no Load Balancing
Technique had used for the efficient execution of tasks in the Cloud. To rectify these gaps, ABC optimized Weighted
Task Load Balancing algorithm has been proposed. The Bee Colony had used to find the best fittest machine for the
efficient execution of tasks in the Cloud and all the tasks can be effectively executed with the best utilization of the
cloud resources.

To begin with the execution, initially, the Cloudsim had initialized by creating the datacenter, broker, machines, and
cloudlets. Tasks list had been created and afterward, the list has been sorted according to the Scheduling algorithm
called as weighted task scheduling and tasks had been prioritized by considering the overall result of the parameters
viz. Length, Priority, and Deadline. Furthermore, apply the ABC optimized algorithm to balance the load on machines
and compute the Fitness value of each virtual machine. If the fitness value is less than the maximum capacity of the
machine than the machine is said to be low-loaded. Then assign the highest weighted tasks to the low-loaded machine,
so that the resources can be well utilized and vice versa.

4.1 Simulation using CloudSim Simulator in Cloud

CloudSim is a platform for modeling and simulation of cloud computing infrastructures and amenities. In the current
scenario, there is a growing demand for IT technologies having repeatable and uncontrollable methodologies for
evaluation of algorithms, applications, and policies prior to the authentic development of cloud products. The
alternative is to use the simulation technique i.e. CloudSim in a cloud environment, where we can perform a number of
actions using the different scenario. Especially, the Internet-based computing, where a high amount of cost is incurred
for the actual development and utilization of resources. There are a number of simulators available for performing the
simulation in the cloud but CloudSim is the unsurpassed simulator to use. Fundamentally Cloudsim has three essential
things i.e. CIS (Cloud Information System), Brokers, Datacenter which is essential to perform all the activities in the
simulator. [21]

Volume 5, Issue 9, September 2017 Page 84


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

4.2 Algorithm Steps

1. Initialize the Cloudsim package by creating the datacenter, broker, machines, and cloudlets.
2. Initialize the machine and Task list.
3. Sorting of Tasks using Weighted Task Scheduling algorithm.
4. Apply Bee Colony Algorithm and calculate the fitness value of the machine and prioritize the machine according
to the fitness value.
5. Map the task to the machine, in which the highest weighted task is mapped to the highest fittest machine.
6. Calculate Load on the machine.
7. This process will be repeated for all tasks.

Figure 3 Flowchart of the proposed methodology


4.3 Parameters of Task Scheduling: The analyzing parameters are as follows: [8, 11, 17-19]

Volume 5, Issue 9, September 2017 Page 85


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

1. Total Processing Time: It is the amount of time taken by the CPU in order to complete the execution of all the
tasks. It is measured using the formula:

2. Total Processing Cost: It is an accounting tactic that traces and heap up direct cost, and allocates the indirect
cost of an executed process which has been accomplished by the utilization of resources. This parameter is
measured using the formula:

3. Average Waiting Time: It is an estimated time spent by all tasks in a ready queue. It is measured using the
formula:

4. Makespan Time: It is the total time that elapses from commencement to end.

5. RESULTS AND DISCUSSION


CloudSim and NetbeansIDE8.0 are used in implementing the proposed methodology. CloudSim is the library that
provides the simulation environment of cloud computing and also provides primary classes describing virtual
machines, data centers, users, and applications. The simulation is exercised for a heterogeneous set of tasks and
machines. In this, several virtual machines are created and they are accessed by the tasks. The main purpose of
CloudSim is to verify the results of the proposed algorithm and perk up the results. The experimental results are
evaluated using the aforementioned parameters.

In the beneath mentioned Table 1, the simulation is performed by taking a varying number of virtual machines and
tasks, the results of the pre-existing and proposed algorithm are generated using the CloudSim simulator. The
Proposed algorithm shows better results in contrast to the pre-existing task scheduling algorithms such as Non-
Dominating Sort and Weighted Task Scheduling.

Table 1 Representing the comparison of Total Processing Time of the proposed with the Existing

ABC Optimized Weighted Task


Weighted Task
[Virtual Machines, Non Dominating Sort (in Scheduling with Load
Scheduling (in
Tasks] milliseconds) Balancing
milliseconds)
(in milliseconds)
[10,100] 787.8198 420.7809 188.2924
[20,200] 2605.9097 1110.3791 609.8652
[30,300] 5161.829 2205.4961 1182.165
[40,400] 8874.0748 3515.8434 1965.0416
[50,500] 12605.9705 4731.8715 2780.6065

Volume 5, Issue 9, September 2017 Page 86


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

Figure 4 Comparison of total processing time of proposed with the existing technique
Above displayed Figure 4 illustrates the contrast between the virtual machines and tasks. From the graph, it is clearly
shown that the proposed algorithm takes a smaller amount time in result to the pre-existing task scheduling algorithm.
Underneath mentioned Table 2, the simulation is practiced by taking a different number of virtual machines and tasks.
In the result, the proposed algorithm has lesser makespan time in weigh against to the pre-existing task scheduling
algorithms such as Non-Dominating Sort and Weighted Task Scheduling.

Table 2 Representing the comparison of Makespan Time of the proposed with the Existing

Weighted Task ABC Optimized Weighted Task


[Virtual Machines, Non Dominating Sort
Scheduling Scheduling with Load Balancing
Tasks] (in milliseconds)
(in milliseconds) (in milliseconds)
[10,100] 68.63 63.63 38.62
[20,200] 232.78 182.87 65.73
[30,300] 475.7 373.41 87.69
[40,400] 813.42 595.99 116.1
[50,500] 1173.74 826.76 127.66

Figure 5 Comparison of Makespan time of proposed with the existing technique

Above illustrated Figure 5 illustrates the relationship between the virtual machines and tasks. From the graph, it is
clearly shown that the proposed algorithm has slighter makespan time in result to the pre-existing task scheduling
algorithm.

Volume 5, Issue 9, September 2017 Page 87


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

In the underneath mentioned Table 3, the estimated cost is based on the pay-per-use service method.
Table 3 Representing the comparison of Total Processing Cost of the proposed with the Existing

ABC Optimized Weighted Task


Weighted Task
[Virtual Machines, Non Dominating Sort (in Scheduling with Load
Scheduling
Tasks] dollar) Balancing
(in dollar)
(in dollar)
[10,100] 17796.235 3695.7849 3334.35
[20,200] 36541.385 11972.135 9011.6999
[30,300] 55286.535 24934.4849 17032.4999
[40,400] 74031.685 42582.835 27395.3999
[50,500] 92776.8349 64917.1849 40101.7499

Figure 6 Comparison of total processing cost of proposed with the existing technique

From the figure 6, it is clear that the processing cost that the user has to pay is slighter in the proposed algorithm than
in the pre-existing algorithm as per the model. Less the processing cost more will be the advantage to the client and
thus increment the overall client's divergence at the cloud service provider.
In Table 4, the proposed algorithm shows the average waiting time of the cloudlets in comparison to the base
algorithm.

Table 4 Representing the comparison of Average Waiting Time of the proposed with the Existing

Weighted Task ABC Optimized Weighted Task


[Virtual Machines, Non Dominating Sort
Scheduling Scheduling with Load Balancing
Tasks] (in milliseconds)
(in milliseconds) (in milliseconds)

[10,100] 0.513249 0.509072 0.227208


[20,200] 0.491278 0.460372 0.254336
[30,300] 0.5030705 0.490901 0.229871
[40,400] 0.500106 0.491593 0.241288
[50,500] 0.507281 0.495415 0.241516

Volume 5, Issue 9, September 2017 Page 88


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

Figure 7 Comparison of average waiting time of proposed with the existing technique

From the Figure 7, it is clearly shown that the tasks have to wait for the less amount of time in contrast to the other two
algorithms and all tasks is executed very quickly.

From the comparison of a four-scheduling parameter, it is observed that the proposed algorithm performs better other
than the two algorithms with less processing time, average waiting time, make spam time and processing cost of cloud
computing system.
6. CONCLUSION
In this paper, the proposed ABC Optimized Weighted Task Scheduling with Load Balancing (ABC-WTSLB) had
comparatively analysed with Weighted Task Scheduling and Non-Dominating sorting algorithms based on parameters
viz. total processing time, make-span time, total processing cost, average waiting time. At different levels of virtual
machines and tasks (10, 100; 50,500), the total processing time of proposed ABC-WTSLB algorithm (188.2 ms; 2780.6
ms) had performed far better than Weighted Task Scheduling (420.7 ms; 4731.8 ms) and Non-Dominating (787.8 ms;
12605.9 ms) algorithms. With the increase of virtual machines and tasks, average waiting time of ABC-WTSLB had
resulted in approximately 2.2 ms whereas Weighted Task Scheduling and Non Dominating algorithms resulted in
approximately 0.50 ms and 0.49 ms. The total processing cost of proposed ABC-WTSLB had been also improved by
approximately 50 % in comparison of Non Dominating sorting and approximately 40 % in comparison of Weighted
Task Scheduling. Our proposed ABC-WTSLB had utilized lesser total processing time, make-span time, total
processing cost and average waiting time in comparison to the Weighted Task Scheduling and Non-Dominating sorting
algorithms.

ACKNOWLEDGMENT
The paper has been originated with the genial help, direction, and support of my department who have helped me in
this work. The author would like to thank all the masses whose encouragement and backing have made the fulfillment
of this work conceivable.

REFERENCES
[1] Ru Jia, An Investigation of Scheduling Policies for Cloud-based Software Systems.
[2] Silva, J.N., Veiga, L., Ferreira, P.: Heuristics for Resource Allocation on Utility Computation Infrastructures.
In:6th IEEE International Workshop on Middleware for Grid Computing, New York (2008)
[3] Peter Mell and Timothy Grance. The NIST Definition of Cloud Computing (Draft), 2011.
[4] L.Wang, G.Laszewski, M.Kunze, J.Tao, "Cloud Computing: A Perspective study," J. New Generation Computing,
pp.1-11, 2010.
[5] Shamsollah Ghanbari, Mohamed Othman, A Priority based Job Scheduling Algorithm in Cloud Computing,
International Conference on Advances Science and Contemporary Engineering 2012 (ICASCE 2012), pp. 778-785

Volume 5, Issue 9, September 2017 Page 89


IPASJ International Journal of Computer Science (IIJCS)
Web Site: http://www.ipasj.org/IIJCS/IIJCS.htm
A Publisher for Research Motivation ........ Email:editoriijcs@ipasj.org
Volume 5, Issue 9, September 2017 ISSN 2321-5992

[6] AV. Karthick, E. Ramaraj, and R. G. Subramanian, "An efficient multi-queue job scheduling for cloud
computing," Proc. World Congress on Computing and Communication Technologies WCCCT 2014, pp. 164166,
2014.
[7] L. Zhang, W. Tong, and S. Lu, Task scheduling of cloud computing based on Improved CHC algorithm, 2014
IEEE International Conference on Audio, Language and Image Processing (ICALIP) vol. 8, no. 8, pp. 574577,
2014.
[8] Y. P. Dave, A. S. Shelat, and D. S. Patel, Various job scheduling algorithms in cloud computing: A survey, 2014
IEEE International Conference on Information Communication & Embedded Systems, no. 978, 2014
[9] Ananth and K. C. Sekaran, "Game-theoretic approaches for job scheduling in cloud computing: A survey," Proc. -
2014 5th IEEE 5th International Conference on Computer and Communication Technology (ICCCT) 2014, pp.
7985, 2014.
[10] Kun Lun Li, W. Jun, S. Jian, and Q. Y. Dong, Improved GEP Algorithm for Task Scheduling in Cloud
Computing, Proc. - 2014 2nd Second International Conference on Advanced Cloud and Big Data, pp. 9399,
2014
[11] A Soni, G. Vishwakarma, Y.K. Jain A Bee Colony based Multi-Objective Load Balancing Technique for Cloud
Computing, International Journal of Computer Application (IEEE), Vol. 114, No. 4, pp-19-25, 2015.
[12] Nidhi Bansal, Amitab Maurya, Tarun Kumar, Manzeet Singh, Shruti Bansal, Cost performance of QoS Driven
task scheduling in Cloud Computing,"2015, Third International Conference on Recent Trends in Computing
(ICRTC 2015), pp. 126-130
[13] A. Gupta, H. S. Bhadauria, A. Singh, and J. C. Patni, A Theoretical Comparison of Job Scheduling Algorithms
in Cloud Computing Environment, 2015 1st International Conference on Next Generation Computing
Technologies (NGCT-2015) pp. 45, 2015.
[14] H. Zhang, Research on Job Security Scheduling Strategy in Cloud Computing Model, 2015 IEEE International
Conference on Intelligent Transportation, Big Data & Smart City, pp. 649652, 2015.
[15] Himani and H. S. Sidhu, Cost-Deadline Based Task Scheduling in Cloud Computing, Proc. -2015 2nd IEEE
Second International Conference on Advances in Computing and Communication Engineering( ICACCE), vol. 6,
pp. 273279, 2015.
[16] T.Mandal and S.Acharyya, Optimal Task Scheduling in Cloud Computing Environment: Meta-Heuristic
Approaches," 2015 Proceedings of International Conference on Electrical Information and Communication
Technology (EICT) pp. 2428, 2015.
[17] A.V. Lakra, D.K. Yadav, Multi-Objective Tasks Scheduling Algorithm for Cloud Computing throughput
optimization, International Conference on Intelligent Computing, Communication & Convergence (Elsevier),
pp.107-113,2015.
[18] A. Thomas et.al., "Credit-Based Scheduling in Cloud Computing Environment", International Conference on
Information and Communication Technologies (Elsevier), pp.913-920, 2015.
[19] L K. Raja, G. Sekar, "An Algorithm for Credit-Based Scheduling in Cloud Computing environment depending
upon Deadline Strategy", International Journal of Current Trends in Engineering and Research, Vol. 2, Issue8,
pp.70-76,2016.
[20] K. Zalavadiya, D. Vaghela, Honey Bee Behavior Load Balancing of Tasks in Cloud Computing, International
Journal of Computer Applications (IEEE), Vol. 139-No. 1, pp. 16-19, 2016.
[21] Dr. Umang Singh, Ayushi Sharma, CloudSim Simulator used for Load Balancing in Cloud Computing,
International Journal of Emerging Technology and Advanced Engineering, Vol. 6 Issue 4, pp. 247-255, 2016.

Volume 5, Issue 9, September 2017 Page 90