An Enhanced Round Robin (ERR) algorithm for

Effective and Efficient Task Scheduling in cloud

environment
Sanaj M S
Research Scholar, Faculty of CSE Dr. Joe Prathap P M
Satyabhama Institute of Science & Department of Information Technology
Technology RMD Engineering College
Chennai, India Chennai, India
[email protected] [email protected]

Abstract—Efficient and Effective scheduling methods can of the major challenges in lieu of resource allocation and has
result in more desired services to the consumers and also can great impact on the performance of cloud computing. Task
improve the performance of a cloud computing environment. scheduling is a challenging process and needs to be improved
The primary job of a task scheduler is to ensure reduction in for ensuring quality and for reducing the cost of computations.
execution time of the tasks and to fulfill maximum resource This should be in such a way that it gives mutual benefit to the
utilization. This paper proposes an Enhanced version of Round consumers as well as the service providers [6],[7]. This paper
Robin algorithm (ERR) for improved performance without proposes an enhanced version of the classical Round Robin
affecting the good features of traditional RR and which can algorithm for ensuring better performance through the
bring out more efficiency. The proposed algorithm is
reduction of waiting time which is a disadvantage of
implemented and tested using CloudSim toolkit and the initial
traditional Round Robin and also to increase the overall
results prove that the average waiting time for the tasks in a
given number of cloudlets is reduced in MRR than the performance in terms of execution time and residual energy
conventional RR in the same conditions. The proposed method [8], [9]. The rest of the paper is presented as follows. The
also outperforms the other existing algorithms such as ACO, GA, second section provides the list of various task schedulers
MPA, Min-Min and PSO in terms of execution time and residue proposed so far. The third section provides the problem of task
energy. scheduling and the fourth section describes the proposed
methodology followed by experimental analysis with results
Keywords—Task Scheduling, Round Robin, FCFS, MCT, in Section 5. The paper concludes with the future remarks in
PSO, CloudSim and Min-Min Section 6.
I. INTRODUCTION II. RELATED WORKS
The cloud computing technology is based on internet for In the section followed, a list on the recent and more
providing services on-demand to the clients from hardware to relevant works carried out for achieving better performance in
network and software applications [1]. The objective is to scheduling the tasks are discussed. The review is presented in
make the clients thinner by sharing resources such as network, a chronological manner for enhancing the presentation.
servers, storages and data centers, Most of the cloud service
providers offer these resources as a pay as you use basis by an In the year 2010, Van den et al [10] proposed an optimum
Service Level Agreement (SLA) and few negotiations in- scheduling algorithm in a hybrid cloud. This cloud was
between the consumer and the provider [2]. The methods for composed of both public and private cloud. The problem here
access of the resources in the network are subjected to follow is that some of the tasks needs be transferred to the public
some universal protocols and networking standards. The cloud from the private in times of peak loads. The Linear
storage and processing of data are usually carried out by the Regressive Method (LRM) was used for the optimization
private clouds that are owned by the organization itself or issue and the results showed increased performance in terms
through a public cloud. Few other cloud models such as hybrid of flexibility and scalability. In 2011, Sindu et al [11]
and multi-cloud are also upcoming in the market [3]. In the proposed couple of algorithms for the task scheduling namely
digital era, cloud computing has become one of the significant the Longest Cloudlet Fastest Processing (LCFP) which tries in
and impressive technology in the market of Internet minimization of the makespan through assigning lengthier
Technology (IT) owing to its multiple advantages such as low tasks to the processing object that has more computational
cost, flexibility, resource sharing , etc. [4]. Also, the efficiency. The second is the Shortest Cloudlet Fastest
sophisticated high end hardware can be avoided which can Processing (SCFP) which performs just the opposite. Here, the
greatly help in cutting the cost for small enterprises. The shorter jobs are mapped to high computational units to avoid
advantage does not limit to reliability and tolerance to faults large queues and minimizes the waiting time. In 2012,
which assures the continuity in services even if some of the Gogulan et al [12] proposed a nature inspired algorithm called
cloud servers are down. the Multiple Pheromone Algorithm (MPA) which is a part of
Ant Colony optimization family. This MPA algorithm
On the advantages on one side, the cloud computing also generates the schedule for a task in a dynamic manner and
has different challenges and hindrances such as the security achieved more performance and Quality of Service (QoS) than
and performance which makes the small companies to hesitate its ancestor Ant Colony Optimization (ACO). In 2013,
adapting the same [5]. The process of task scheduling is one Kaleeswaran et al [13] implemented the Genetic Algorithm

978-1-7281-6453-3/20/$31.00 ©2020 IEEE

107
Authorized licensed use limited to: University of Glasgow. Downloaded on November 02,2020 at 07:13:05 UTC from IEEE Xplore. Restrictions apply.
(GA ) for solving the issue of uncertainty among the tasks that For a set of given tasks k such as T= {T1, T2, . . . .Tk}
arrive in various times into a cloud. The problem creates a with N resources such as R= {R1,R2, . . . .Rk} in the present
huge load on the Virtual Machines (VMs). The proposed cloud. The resources here refer to the VM. The main objective
method used the concept of dynamic scheduling where a of the task scheduler is to minimize the processing time and to
queue is maintained and the scheduler does the job of sorting maximize the throughput still maintaining a good QoS and
them based on the memory and computational requirements. without compromising the SLA. This can be represented as in
Eq(1,2)
In 2014, Agarwal et al [14] proposed a novel method for
task scheduling. The Generalized Priority Algorithm (GPA)
𝑀𝑖𝑛 (𝑃𝑇 ) = 𝑀𝑖𝑛 (∑𝐾𝑖 = 1 ( 𝑇𝑗 ∗ 𝐿𝑒𝑛𝑔𝑡ℎ / 𝑉𝑀𝑚𝑖) (1)
was proposed and was experimented versus the RR and the
FCFS Schemes. The number of VMs was dynamically
changed and the results are monitored using CloudSim. The 𝑀𝑖𝑛 (𝑊𝑇) = 𝑀𝑖𝑛 ((𝑇𝑖. 𝑊𝑇) (2)
results proved better efficiency than RR and First Come First
Serve (FCFS) methods. In 2015, Lakra et al [15] introduced Where PT is the processing time and Ti is the length of the
the Multi task objective scheduler in-order to correctly map task. VMmi is the total no of instructions per second that
the tasks with the VMs. Unlike the single criteria methods measures the speed of CPU. The VM allocation between the
which take the execution time alone, the proposed method tasks is to reduce the workload. The time complexity is
takes additional performance parameter such as cost, user considered to be bit complex as there may be many factors that
bandwidth, etc. The results obtained through the CloudSim influence the parameters such as completion ratio , cost,
tool proves that the proposed method gives more throughout energy consumption, etc. On the contrary, the task scheduling
without being deviated from the SLA. In 2016, Hammad et al problem is non-polynomial and hence need an optimization
[16] proposed the TS-GA a type of genetic algorithm for technique to solve the issue.
solving the scheduling issue of the cloud. The method aims to IV. PROPOSED METHODOLOGY
minimize the execution time and to maximize the used
resources. The results prove to be efficient in terms of speedup The Round Robin algorithm is considered to perform well
and resource sharing than traditional GA and RR algorithms. in case of many CPU scheduling problems and have achieved
In 2017, Madni et al [17] made a comparative analysis of the great performance through fair allocation policies which are
popular heuristic methods based on top 6 parameters. The based on the time frames in each of the processes. The
Min-Min method was found to be effective than all other processes that are running are forced and preempted from the
methods. The heuristic property was then taken as a base and CPU into a queue and hence avoids starvation in the system
many other child algorithms were presented. by making a long wait time for the processes to be allotted
CPU. If there are N tasks, then every task will be allotted 1/N
In 2018, D. Wu et al [18] introduced a task scheduler based of the VM’s processing time and the waiting time will not be
on the Particle Swarm Optimization (PSO) algorithm for more than (n-1)*QT and the overhead of scheduling will be of
obtaining more efficiency. The PSO are implemented to solve the order of o(1). But, the Cloud environment are subjected to
the optimization issue in the task scheduling. The face many changes while the tasks wait for a longer time in
experimental results show that the execution time is reduced the queue and fixed time slicing method cannot assure best
and can be used for the complex problems of scheduler system performance. The time of wait and the ratio of the
optimization. In 2019, Yamin Thet et al [19] introduced the context switching need to be handled cautiously.
task schedulers of the VMs in the cloud. The computations are
then carried out in the separate space thus reducing memory Taking the facts above into consideration, an enhanced RR
and allow more space for large tasks. The experimental results was proposed to solve the bottleneck of the traditional RR.
through the CloudSim proves that the execution time of the The traditional RR is modified for enabling it to make
proposed method is very less when compared to that of the dynamic decisions by making suitable adjustments to the time
others. quantum based on a given situation which can give positive
signs in terms of response time. Here, ERR is presented and a
From the review, it is evident that there is nothing like a comparison is made between the RR and ERR on same
universal solution to the task scheduling problem. Some experimental conditions. The task scheduling concept based
methods were implemented for obtaining more efficiency and on the dynamic allocation of time quantum is given in Figure
others were for less time complexity. Some are proposed to 1. Here, the time slices are calculated in a dynamic manner
reduce makespan and others for increasing speed up and few and hence are subjected to vary based on the task considered.
for avoiding starvations. It is left to the need of the hour and The time quantum as shown in figure will be different for
the cloud environment to implement task schedulers for which different tasks
it needs to be deployed.
III. PROBLEM STATEMENT
Task scheduling in cloud computing is concerned with the
process of assigning user task to the resources that are
available in a way such that there is a maximum utilization of T1-4
the resource and minimized time complexity [20]. It is the T2-8 T3-11 Tn - 18
process of mapping the continuous users’ task to the various
resources in the cloud computing environment. This is a Fig 1- Dynamic Task Allocation
problem of optimization as the scheduler is expected to find in
an optimal way in which the task-VM mapping could be done
in a best way such that the time involved in allocation and Fig. 1. Dynamic Task Allocation
response are minimum with minimal make span.

108
Authorized licensed use limited to: University of Glasgow. Downloaded on November 02,2020 at 07:13:05 UTC from IEEE Xplore. Restrictions apply.
 8 12 9.08 0.04
9 16 12.18 0.09
10 21 17.16 0.12
The pseudo code of the ERR is given below:
ALGORITHM
IP : Cloud(tasks) , VM
OP : map each task to VM
Steps:
NO_CL : List(cloudlet).Sizeof()
No_VM: VM.Sizeof()
I= 0
FOR J=0 ; No(CL) do
CL.list.getlist ();
Index = Index +1 ; No_VM
VM.get(indices)
Stage(transfer_time(CL, VM,Index)
Out.trans_time(Cl , V)
Ex. Exec_time(Cl.V)
IF Cl.AT+Stage.in+exec+sta<=Cl.Dl THEN
Send_job(Cl.V)
Update (V) Else
Drop (CL)
Failed(tasks)
Stop

The ERR proposed above is a minor variant of the Fig. 2. Comparison of AWT
traditional RR as it considers the time slices in a dynamic
manner for each task that enters into the queue. The time slices TABLE 2 – COMPARISON OF PERFORMANCE
are assigned based on the Mean Bursting time of all the
waiting tasks in the queue. Couple of memory registers
namely the SREG and AREG is used for achieving this. The Methods Evaluation Measures
former stores the total burst time required and the latter Execution Time Residual Energy
denotes the average bursting time. In the beginning, the first ACO 16.12 18.15
task is allotted to a VM and it takes all the time. Then, the GA 15.26 17.62
scheduler does the job of calculating the time slice required MPA 14.85 19.28
for the entire incoming task thereafter. Every task when FCFS 17.12 13.95
allotted a VM will be running for a certain time quantum and PSO 12.15 16.62
when it lapses, it is either moved back to the queue and the MRR 08.18 10.18
completed task will be removed out of the queue. The
scheduler hence adjusts the register values by subtracting the
time quantum from the SREG and by adding the same to the Table 2 shows the comparative analysis of the
other register AREG. performance of other scheduling methods with that of the
ERR and the proposed method is found to be better in terms
V. EXPERIMENTAL SETUP AND RESULTS of Execution time per 100 cloudlets considered and Residual
The experiments were conducted in CloudSim energy. The same is depicted in Figure 3.
environment which provides the all functionalities as that of a
real cloud environment. Specific scheduling mechanisms can
also be simulated with customizable options. The efficiency
and the performance can also be tested here. The proposed
MMR is tested in the toolkit with different values for VM and
compared with that of traditional RR in the same VM
environment. The no of VMs were altered exponentially and
simultaneously the cloudlets (tasks) were also increased. The
results are discussed below.
Table 1 show that the AWT of ERR is found to be much
lesser than that of the RR. The same is depicted in the figure
2. The size of the time quantum is increased up to 5 times the
initial value.

TABLE 1 – COMPARISON OF AWT

No of tasks AWT in RR AWT in Relative Error
ERR
5 5 4.21 0.16 Fig. 3. Performance Comparison

6 7 6.11 0.12
7 9 7.62 0.12

109
Authorized licensed use limited to: University of Glasgow. Downloaded on November 02,2020 at 07:13:05 UTC from IEEE Xplore. Restrictions apply.
 VI. CONCLUSION [17] Madni S , Latif A and Usman M J ," Performance Comparison of
Heuristic algorithms for task scheduling in IaaS cloud computing
Task scheduling is a challenging process in cloud environment", Plosone, Vol 3(2) ,2017, pp 36-47.
environment and are directly responsible for the performance. [18] D. Wu, "Cloud Computing Task Scheduling Policy Based on Improved
The task scheduling appears to be the most mandates to Particle Swarm Optimization," 2018 International Conference on
achieve various performance measures. This paper proposes Virtual Reality and Intelligent Systems (ICVRIS), Changsha, 2018, pp.
99-101.
an enhanced Round Robin Algorithm for reducing the waiting
time of the tasks in the ready queue as in traditional RR. The [19] Y. T. H. Hlaing and T. T. Yee, "Static Independent Task Scheduling
on Virtualized Servers in Cloud Computing Environment," 2019
results show that the proposed method produces less Average International Conference on Advanced Information Technologies
waiting time for tasks and also found to be more effective than (ICAIT), Yangon, Myanmar, 2019, pp. 55-59.
the other scheduling algorithms in terms of residual energy [20] P. Zhang and M. Zhou, "Dynamic Cloud Task Scheduling Based on a
and execution time. Two-Stage Strategy," in IEEE Transactions on Automation Science
and Engineering, vol. 15, no. 2, pp. 772-783, April 2018.
REFERENCES
[1] Priyanshu Srivastava and Rizwan Khan, “A Review Paper on Cloud
Computing,” International Journals of Advanced Research in
Computer Science and Software Engineering, vol. 8. , issue 6, pp 17-
20, June 2018.
[2] H. Al-Zoubi, "Efficient Task Scheduling for Applications on Clouds,"
2019 6th IEEE International Conference on Cyber Security and Cloud
Computing (CSCloud)/ 2019 5th IEEE International Conference on
Edge Computing and Scalable Cloud (EdgeCom), Paris, France, 2019,
pp. 10-13.
[3] W. Z. Jiang and Z. Q. Sheng, "A New Task Scheduling Algorithm in
Hybrid Cloud Environment," 2012 International Conference on Cloud
and Service Computing, Shanghai, 2012, pp. 45-49.
[4] X. Zhu, L. T. Yang, H. Chen, J. Wang, S. Yin and X. Liu, "Real-Time
Tasks Oriented Energy-Aware Scheduling in Virtualized Clouds,"
IEEE Transactions on Cloud Computing, vol. 2, no. 2, pp. 168-180,
April-June 2014.
[5] J. Xue, L. Li, S. Zhao and L. Jiao, "A Study of Task Scheduling Based
on Differential Evolution Algorithm in Cloud Computing," 2014
International Conference on Computational Intelligence and
Communication Networks, Bhopal, 2014, pp. 637-640.
[6] S. Selvarani and G. S. Sadhasivam, "Improved cost-based algorithm
for task scheduling in cloud computing," 2010 IEEE International
Conference on Computational Intelligence and Computing Research,
Coimbatore, 2010, pp. 1-5.
[7] Sanaj MS and Joe Prathap P M, “Profit maximization based task
scheduling in hybrid clouds using whale optimization technique,”
Information Security Journal: A Global Perspective, vol. 29, 2020-
Issue 4, 2020, pp 155-168.
[8] Sanaj MS and Joe Prathap P M, “Nature inspired chaotic squirrel search
algorithm (CSSA) for multi objective task scheduling in an IaaS cloud
computing atmosphere,” Engineering Science and Technology, an
Internationall Journal, https://doi.org/10.1016/j.jestch.2019.11.002,
November 2019.
[9] G. Shobana, M. Geetha and R. C. Suganthe, "Nature inspired
preemptive task scheduling for load balancing in cloud datacenter,"
International Conference on Information Communication and
Embedded Systems (ICICES2014), Chennai, 2014, pp. 1-6.
[10] Van den,"Cost-optimal scheduler for hybrid cloud platform, IEEE 3rd
International Conference on cloud computing, 2010, pp. 245-256.
[11] Sindu M, "Efficient task Scheduler for cloud computing," High
performance computing, Springer, Hiddenberg, Berlin, 2011, pp. 79-
83.
[12] Gogulan R, Kumar, “Multi-Pheromone algorithm for scheduling tasks
in cloud using different QoS parameters", Int. J. Comput. Sci, Vol 12 ,
Issue 3, 2012, pp 48-56.
[13] A Kaleeswaran , V Ramasamy and P Vivekanandan, "Dynamic
Scheduling of data using genetic algorithms in cloud computing",
International Journal Advances in Engineering &
Technology,Vol.5,Issue 2, 2013, pp. 327-334.
[14] Agarwal, Jain D, "Efficient Optimal Algorithm of Task Scheduling in
Cloud Computing Environment", axriv : reprint : 1504:2076,2014.
[15] Lakra A and D K Yadav, "Multi-Objective Tasks Scheduling
Algorithms for Cloud Computing throughput optimization", Procedia ,
CS 49 , 2015, pp 452-461.
[16] Hammad and Omara F, "GA-Based Task Scheduler for cloud
environment", Int.J.ADv.Cpmp.Sci , 7(4),2016, pp 551-557.

110
Authorized licensed use limited to: University of Glasgow. Downloaded on November 02,2020 at 07:13:05 UTC from IEEE Xplore. Restrictions apply.