International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)

Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com

Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 151


Abstract: Cloud computing is a distributed computing
model in which enables developers to automatically deploy
applications during job allocation and storage distribution.
Cloud computing intends to share a pool of virtualized
computer resources and equipment's of computation, storage,
information and knowledge for scientific researches over the
Internet. Job scheduling (allocation hardware resources to
the tasks) is one of the major and challenging issues in cloud
computing. A good scheduling policy also helps in proper and
efficient utilization of Virtual Machines (VMs). Job
scheduling algorithm is an NP-complete problem which
places an important role in cloud computing. This paper
surveys the existing different type of job scheduling
algorithms in cloud computing environment and summarizes
some method to improve the performance.

Keywords: Survey, Cloud Computing, Job Scheduling.

1. Introduction
Cloud Computing is an essential ingredient of advanced
computing systems. Computing concepts, technology and
architectures have developed and consolidated in the last
decades. Many aspects are subject to technological
evolution and revolution. Cloud Computing is a
computing technology that is rapidly consolidating itself
as the next step in the development and deployment of
increasing the number of distributed application [1].
To gain the maximum benefit from cloud computing,
developers must design mechanisms that optimize the use
of architectural and deployment paradigms. The role of
Virtual Machines (VMs) has emerged as an important
issue because, through virtualization technology, it makes
cloud computing infrastructures to be scalable. Therefore
developing on optimal scheduling of virtual machines is
an important issue.
The cloud computing architecture has three layers [2], for
the software which require on demand services over
Internet. Figure 1 shows different layers of cloud
architecture.
1. Infrastructure as a Service (IaaS): This layer delivers
hardware components (like server and storage) and
software as services.
2. Platform as a Service (PaaS): Cloud application
developers are the users of this layer. Automatic scaling,
load balancing and integration with other services (e.g.
email services) are the major benefits to cloud application
developer.
3. Software as a Service (SaaS): This layer hosts the
software and provide to the customer through Internet. It
reduces the purchase and maintenance cost of the
customer.
The main Purpose is to schedule tasks to the Virtual
Machines (VMs) in accordance with adaptable time,
which involves finding out a proper sequence in which
tasks can be executed under transaction logic constraints
[3]. The job scheduling of cloud computing is a
challenge. To take up this challenge we review the
number of efficiently job scheduling algorithms. It aims at
an optimal job scheduling by assigning end user task.


Figure 1 Cloud computing service

The rest of the paper is organized as follows. In next
section Literature Survey about different scheduling
algorithms of Virtual machine in cloud are discussed.
Section 3 describes Existing Scheduling Algorithm in
Cloud Computing. Section 4 discusses The Proposed
Scheduling Algorithm. Section 5 discusses Experimental
Setup and Results are analyzed. Conclusions are
discussed in section 6.

2. Task Scheduling in Cloud Computing
Job Scheduling of cloud computing refers to dispatch the
computing tasks to resource pooling between different
resource users according to certain rules of resource use
under a given cloud circumstances. At present there is not
a uniform standard for job scheduling in cloud
computing. Resource management and job scheduling are
the key technologies of cloud computing that plays a vital
role in an efficient cloud resource management.

3. Existing Scheduling Policy
The following task scheduling algorithms are presently
established in the cloud environments and these
algorithms have been summarized in table 1 with
scheduling parameters.
3.1. Ant Colony Optimization (ACO)-inspired:
A new Cloud scheduler based on Ant Colony
Optimization is the one presented by Cristian Mateos and
et.al [4]. The goal of our scheduler is to minimize the
Survey on Job Scheduling algorithms in Cloud
Computing

Fazel Mohammadi
1
, Dr. Shahram Jamali
2
and Masoud Bekravi
3


1
Department of computer, Science and Research branch, Islamic Azad University, Ardabil, Iran
2
Associate Professor, University of Mohaghegh Ardabili, Ardabil, Iran
3
Department of Computer Engineering, Ardabil Branch, Islamic Azad University, Ardabil, Iran
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 152


weighted flowtime of a set of PSE jobs, while also
minimizing Makespan when using a Cloud.
In the ACO algorithm, the load is calculated on each host
taking into account the CPU utilization made by all the
VMs that are executing on each host. This metric is
useful for an ant to choose the least loaded host to allocate
its VM.
Parameter Sweep Experiments (PSE) is a type of
numerical simulation that involves running a large
number of independent jobs and typically requires a lot of
computing power. These jobs must be efficiently
processed in the different computing resources of a
distributed environment such as the ones provided by
Cloud. Consequently, job scheduling in this context
indeed plays a fundamental role.
In this algorithm, Makespan and flowtime are evaluated.
Evaluation results of this metrics show that ACO
performance better than two other (Random and Best
effort) algorithms.

3.2. Berger model:
Baomin Xu and et.al [5] proposed for the first time an
algorithm of job scheduling based on Berger model in
cloud environment. The Berger model of distributive
justice is based on expectation states. It is a series of
distribution theories of social wealth.
Expectation states formed by a series of theories are used
to study actors and evaluate the impact of their behavior.
Brief speaking, expectation states theories are to study the
follow two issues. First, actor how to generate
expectations of itself and other individuals according to
the information (such as status, reward, and performance
differences) around the world; Secondly, these
expectations how to affect the behavior (such as
participatory, and decision-making influence) of actors
and others. Expectation states theories have been
expanded and applied widely. State value theory of
distribution justice is an important theoretical basis of the
paper. It described that allocator how to use referential
comparisons to establish the expectation for reward
allocation. The expectations are used to evaluate the
justice or injustice of distribution in a variety of
circumstances.
Through the expansion of CloudSim platform, job
scheduling algorithm based on Berger model is
implemented. The validity of the algorithm is verified on
the extended simulation platform. By comparing of
simulation results with the optimal completion time
algorithm, the proposed algorithm in this paper is
effective implementation of user tasks, and with better
fairness.
3.3. Dynamic level scheduling (DLS) algorithm:
Wei Wang and et.al extend the traditional DLS algorithm
by considering trustworthiness of resource nodes [6]. This
algorithm meets the requirement of user jobs in trust, and
makes jobs scheduling based on directed acyclic graph
(DAG) more reasonable. The main contribution of this
study to scheduling systems is that it extends the
traditional formulation of the scheduling problem so that
both execution time and reliability of applications are
simultaneously accounted for.
Evaluation the trustworthiness of machines in Cloud
environment by proposed algorithm shows decrease the
failure probability of the task assignments, and assurance
of the execution of tasks in a security environment.
3.4. Min-Min Algorithm:
The main idea of the Min-Min algorithm is as quickly as
possible to dispatch each task to virtual machines as
resources which can complete the task in the shortest
possible time [7].
Min-Min algorithm will execute short jobs in parallel and
the long jobs will follow the short jobs. The shortcoming
of this algorithm is the short jobs scheduled first, until the
machines are leisure to schedule and execute long jobs.
Min-min can cause both the whole batch jobs executed
time get longer and unbalanced load. Even long jobs
cannot be executed. Compared with the traditional Min-
min algorithm, improved algorithm adds the three
constraints (quality of service, the dynamic priority model
and the cost of service) strategy which can change this
condition.
The experimental results of improved Min-Min algorithm
show it can increase resource utilization rate, long tasks
can execute at reasonable time and meet users
requirements.
3.5. Particle Swarm Optimization (PSO) Algorithm:
Particle Swarm Optimization (PSO) as a meta-heuristics
method is a self-adaptive global search based
optimization technique introduced by Kennedy and
Eberhart [8]. The PSO algorithm is alike to other
population-based algorithms like Genetic algorithms
(GA) but, there is no direct recombination of individuals
of the population [9].
The PSO algorithm focuses on minimizing the total cost
of computation of an application workflow.
As a measure of performance, Authors used cost for
complete execution of application as a metric. The
objective is to minimize the total cost of execution of
application workflows on Cloud computing environments.
Results show that PSO based task-resource mapping can
achieve at least three times cost savings as compared to
Best Resource Selection (BRS) based mapping for our
application workflow. In addition, PSO balances the load
on compute resources by distributing tasks to available
resources.
3.6. Priority based job scheduling (PJSC) Algorithm:
Priority of jobs is an important issue in scheduling
because some jobs should be serviced earlier than other
those jobs cant stay for a long time in a system.
In cloud computing, the main objective of this algorithm
is to propose a new priority based job scheduling
algorithm called PJSC. The proposed algorithm is based
on the theory of Analytical Hierarchy Process (AHP). The
Analytical Hierarchy Process is a multi-criteria decision-
making (MCDM) and multi-attribute decision-making
(MCDM) model [10].
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 153


The PJSC algorithm is based on multiple criteria decision
making model.
The PJSC algorithm provided a discussion about some
issues such as complexity, consistency and finish time.
Evaluation result of this algorithm has reasonable
complexity, also it decrease finish time (Makespan).
3.7. Round Robin Algorithm:
The Round Robin algorithm mainly focuses on
distributing the load equally to all the resources [11].
Using this algorithm, the broker allocates one VM to a
node in a cyclic manner. The round robin scheduling in
the cloud computing is very similar to the round robin
scheduling used in the process scheduling. The scheduler
starts with a node and moves on to the next node, after a
VM is assigned to that node. This is repeated until all the
nodes have been allocated at least one VM and then the
scheduler returns to the first node again. Hence, in this
case, the scheduler does not wait for the exhaustion of the
resources of a node before moving on to the next.
Although round robin algorithms are based on simple
rule, more load is conceived on servers and thus
unbalancing the traffic.
Result of Round Robin algorithm shows better response
time and load balancing as compared to the other
algorithms.


Table 1. Scheduling Metrics considered by existing task Scheduling Algorithms
Paper
E
x
e
c
u
t
i
o
n

T
i
m
e

R
e
s
p
o
n
s
e

T
i
m
e

T
r
u
s
t

M
a
k
e
s
p
a
n

U
t
i
l
i
z
a
t
i
o
n


F
l
o
w
t
i
m
e

C
o
s
t

S
u
c
c
e
s
s
f
u
l

e
x
e
c
u
t
i
o
n

R
e
l
i
a
b
i
l
i
t
y

F
a
i
r
n
e
s
s

L
o
a
d

B
a
l
a
n
c
i
n
g


R
e
a
s
o
n
a
b
l
e

C
o
m
p
l
e
x
i
t
y


An ACO-inspired algorithm for
minimizing weighted flowtime in cloud-
based parameter sweep experiments [4]

J ob scheduling algorithmbased on Berger
model in cloud environment [5]

Cloud-DLS: Dynamic trusted scheduling
for Cloud computing [6]

An Improved Min-Min Algorithmin Cloud
Computing [7]

A Particle SwarmOptimization (PSO) -
based Heuristic for Scheduling Workflow
Applications in Cloud Computing
Environments [8]

A Priority based J ob Scheduling Algorithm
in Cloud Computing [10]

Analysis of variants in Round Robin
Algorithms for load balancing in Cloud
Computing [11]

Load Balancing in Cloud Computing using
Stochastic Hill Climbing-A Soft
Computing Approach [12]


3.8. Stochastic Hill Climbing-A Soft Computing
Approach:
A local optimization approach Stochastic Hill climbing is
used for allocation of incoming jobs to the servers or
virtual machines (VMs) [12].
A variant of Hill Climbing algorithm Stochastic Hill
Climbing (SHC) is one of the incomplete approaches for
solving such optimization problems. The Incomplete
methods may not guarantee correct answers for all inputs.
Rather these methods find satisfying assignments for
solvable problems with high probability.
With the scenario and configuration, overall average
response time (RT) is calculated. The results show that in
most of the case Stochastic Hill Climbing (SHC) decrease
response time.
4. Our Contributions
A scheduling framework can be implemented by using
different parameters. Good scheduling framework should
include the following specifications. It must focus on:
Load balancing and Energy efficiency of the data
centers and virtual machines (VMs)
Quality of Service parameters calculated by the user
which contain execution time, cost and so on
It should satisfy the security features.
Fairness resource allocation places a vital role in
scheduling
Considering all the metrics in a single scheduling
framework is not a possible solution, hence it increases
International Journal of EmergingTrends & Technology in Computer Science(IJETTCS)
Web Site: www.ijettcs.org Email: editor@ijettcs.org, editorijettcs@gmail.com
Volume 3, Issue 2, March April 2014 ISSN 2278-6856


Volume 3, Issue 2 March April 2014 Page 154


the complexity of the design; the scheduling framework
can be implemented.

5. Conclusion
Cloud computing is one of the user oriented technology in
which user faces a pool of virtualized computer resources.
In this paper we survey various existing scheduling
algorithms in cloud computing. Table 1 is shown for
additional reference. Since cloud computing is in infancy
state, a scheduling framework should be implemented to
improve the user acquiescence along with the service
providers. The scheduling metrics can be coupled to
prepare a framework for recourse allocation and
scheduling in cloud computing. The scheduling
framework should consider the user input limitations
(deadlines, performance issues, execution cost,
transmission cost, energy efficiency, Load Balancing, and
Makespan) and so on.

References
[1] J udeans and S. Ghemawat, MapReduce: simplified data
processing on large clusters, Sixth Symposium on
Operating SystemDesign and Implementation (OSDI04),
Dec. 2004, pp. 1-13.
[2] Humayoun Israr, (2013), ARCHITECTURE LEVEL
MAPPING OF CLOUD COMPUTING WITH GRID
COMPUTING, International J ournal of Engineering
Sciences & Emerging Technologies, Volume 5, Issue 1, pp.
7-11.
[3] XIAO Zhi-J iao, CHANG Hui-You, YI Yang, (2007), An
Optimization Method of Workflow Dynamic Scheduling
Based on Heuristic GA, Computer Science, Vo1.34 No.2.
[4] Cristian Mateos, Elina Pacini & Carlos Garc Garino,
(2013), An ACO-inspired algorithm for minimizing
weighted flowtime in cloud-based parameter sweep
experiments.
[5] Baomin Xu, Chunyan Zhao, Enzhao Hu, Bin Hu, (2011),
J ob scheduling algorithmbased on Berger model in cloud
environment, Advances in Engineering Software 42, PP.
419425.
[6] Wei Wang, Guosun Zeng, Daizhong Tang and J ing Yao,
(2012), Cloud-DLS: Dynamic trusted scheduling for Cloud
computing, Expert Systems with Applications 39, PP.
23212329.
[7] Gang Liu, J ing Li and J ianchao Xu, (2013), An Improved
Min-Min Algorithmin Cloud Computing, Department of
Computer Science and Engineering, Changchun University
of Technology, China.
[8] Suraj Pandey1, LinlinWu1, Siddeswara Mayura Guru2 &
Rajkumar Buyya, A Particle Swarm Optimization-based
Heuristic for Scheduling Workflow Applications in Cloud
Computing Environments, Cloud Computing and
Distributed Systems Laboratory, Department of Computer
Science and Software Engineering, The University of
Melbourne, Australia.
[9] J . Kennedy and R. Eberhart, (1995), Particle swarms
optimization In IEEE International Conference on Neural
Networks, volume 4, pages 19421948.
[10] Shamsollah Ghanbaria & Mohamed Othmana, (2012), A
Priority based J ob Scheduling Algorithm in Cloud
Computing, Procedia Engineering 50, and PP. 778 785.
[11] Pooja Samal and Pranati Mishra, (2013), "Analysis of
variants in Round Robin Algorithms for load balancing in
Cloud Computing", International J ournal of Computer
Science and Information Technologies, pp. 416-419, Vol. 4
(3).
[12] Brototi Mondala, Kousik Dasgupta & Paramartha Duttab,
(2012), Load Balancing in Cloud Computing using
Stochastic Hill Climbing-A Soft Computing Approach,
Procedia Technology 4, PP. 783 789.



AUTHORS

Fazel Mohammadi is currently a M.Sc.
student in the Department of Computer in the
Science and Research branch, Islamic Azad
University, Ardabil, Iran. His research
interests are in the areas of distributed systems with
emphasis on scheduling algorithms for grids and clouds,
cloud computing and machine learning. He is a teacher in
Computer Science at technical high school and technical
college.

Shahram Jamali is an associate professor
leading the Computer Networking Group at
the Department of Engineering, University of
Mohaghegh Ardabili. He teaches on computer
networks, network security, computer architecture and
computer systems performance evaluation. Dr. J amali
received his M.Sc. and Ph.D. degree from the Dept. of
Computer Engineering, Iran University of Science and
Technology in 2001 and 2007, respectively. Since 2008,
he is with Department of Computer Engineering,
University of Mohaghegh Ardabil and has published
more than 80 conference and journal papers. He also
serves as editor member for some international journals

Masoud Bekravi received his B.Sc. and M.Sc.
degrees in Electrical Engineering and
Computer Engineering in 2004 (Islamic Azad
university-Ardabil branch) and 2007 (Islamic
Azad university-Arak branch) respectively. Now he is
faculty member in the CE department, Islamic Azad
University Ardabil branch. His research interests are
power system, artificial intelligence, swarm robotics and
Computer systems.