ArticlePDF Available

The Study of Genetic Algorithm-based Task Scheduling for Cloud Computing

Authors:

Abstract and Figures

Task scheduling is an important and challenging issue of Cloud computing. Existing solutions to task scheduling problems are unsuitable for Cloud computing because they only focus on a specific purpose like the minimization of execution time or workload and do not use characteristics of Cloud computing for task scheduling. A task scheduler in Cloud computing has to satisfy cloud users with the agreed QoS and improve profits of cloud providers. In order to solve task scheduling problems in Cloud computing, this paper proposes a task scheduling model based on the genetic algorithm. In the proposed model, the task scheduler calls the GA scheduling function every task scheduling cycle. This function creates a set of task schedules and evaluates the quality of each task schedule with user satisfaction and virtual machine availability. The function iterates genetic operations to make an optimal task schedule. Experimental results show effectiveness and efficiency of the genetic algorithm-based task scheduling model in comparison with existing task scheduling models, which are the round-robin task scheduling model, the load index-based task scheduling model, and the ABC based task scheduling model.
Content may be subject to copyright.
International Journal of Control and Automation
Vol. 5, No. 4, December, 2012
157
The Study of Genetic Algorithm-based Task Scheduling
for Cloud Computing
Sung Ho Jang, Tae Young Kim, Jae Kwon Kim and Jong Sik Lee
School of Information Engineering
Inha University #253, YongHyun-Dong, Nam-Ku
Incheon 402-751, Republic of Korea
ho7809@hanmail.net, silverwild@gmail.com, jaekwonkorea@naver.com,
jslee@inha.ac.kr
Abstract
Task scheduling is an important and challenging issue of Cloud computing. Existing
solutions to task scheduling problems are unsuitable for Cloud computing because they only
focus on a specific purpose like the minimization of execution time or workload and do not
use characteristics of Cloud computing for task scheduling. A task scheduler in Cloud
computing has to satisfy cloud users with the agreed QoS and improve profits of cloud
providers. In order to solve task scheduling problems in Cloud computing, this paper
proposes a task scheduling model based on the genetic algorithm. In the proposed model, the
task scheduler calls the GA scheduling function every task scheduling cycle. This function
creates a set of task schedules and evaluates the quality of each task schedule with user
satisfaction and virtual machine availability. The function iterates genetic operations to make
an optimal task schedule. Experimental results show effectiveness and efficiency of the
genetic algorithm-based task scheduling model in comparison with existing task scheduling
models, which are the round-robin task scheduling model, the load index-based task
scheduling model, and the ABC based task scheduling model.
Keywords: Task Scheduling, Genetic Algorithm, Cloud Computing
1. Introduction
With the development of system virtualization and Internet technologies, Cloud computing
has emerged as a new computing platform. Cloud computing is to provide virtualized IT
resources as cloud services by using the Internet technology [1]. In Cloud computing, a cloud
user commits to an agreement called Service Level Agreement (SLA) [2] with a cloud
provider. The cloud user utilizes IT resources like storage and server as a service and pays for
the service. A cloud provider constructs a computing system called cloud, which consists of
several virtual machines interconnected, and makes profits by processing tasks from users on
the computing system. Therefore, how to allocate tasks to virtual machines efficiently is an
important and challenging issue in Cloud computing.
The main goal of Cloud computing is to satisfy cloud users with the agreed QoS and
improve profits of cloud providers [3]. A task scheduling model is necessary to achieve the
main goal. There are various task scheduling models [4, 5, 6] developed by a lot of
researchers to utilize computing resource in distributed computing environments, but the
existing models are unsuitable for Cloud computing. Because, the existing models mainly
focus on improvement of system performance. Most of the task scheduling models for Cluster
computing tried to minimize the completion time of a batch of tasks. Task scheduling models
International Journal of Control and Automation
Vol. 5, No. 4, December, 2012
158
for Grid computing, the forerunner of Cloud computing, aimed at improvement of specific
performance metrics like computing speed and storage availability. Besides, user satisfaction
and provider's profit must be considered to solve task scheduling problems in Cloud
computing. In other words, a scheduler makes task schedules, which meet the QoS agreed on
a SLA between cloud users and cloud providers and augment profits of cloud providers.
Rest of this paper is as follows: In Section 2, we propose the GA-based task scheduling
model for Cloud computing and describes how to design elements of the GA task scheduling
function. Section 3 shows the configuration of simulation and brief results. Finally,
conclusion is given in Section 4.
2. GA-based Task Scheduling for Cloud Computing
In this section, we propose the GA-based task scheduling model and illustrate its design.
We also explain how to make an optimal task schedule and compose elements of the GA
scheduling function. As mentioned in the previous section, a cloud user reaches a SLA with a
cloud provider to process a task. A SLA document includes user requirements like time and
budgetary constraints of the task, which indicate acceptable deadline and payable budget of
the cloud user. QoS attributes like response time and throughput can be comprised in a SLA
document besides time and budgetary constraints [2]. A cloud provider has to consider user
requirements and virtual machine information before allocating tasks from to virtual
machines.
2.1. Encoding and Initiation
A chromosome chk indicates the task allocation information, i.e. a task schedule. k is
from 1 to z, which denotes the number of chromosomes in a population. In
consideration of user satisfaction and providers profit, the task scheduler determines
where to allocate each task every scheduling cycle. A chromosome chk consists of α[i]
and β[i], which indicate the information of task processing and virtual machine
allocation. The length of a chromosome is the same as the number of inputted tasks.
The encoding operation to express a task schedule as a chromosome is as follows. A set of
inputted tasks is sorted before the creation of chromosomes. Unlike existing distributed
computing environments, cloud users pay for computing services in person. Accordingly,
tasks received from cloud users with high cost have to be allocated to virtual machines faster
than others. In this paper, we classify takes into four groups by time and budgetary constraints
as shown in Figure 1.
Figure 1. Classification of Tasks by Time and Budgetary Constraints
International Journal of Control and Automation
Vol. 5, No. 4, December, 2012
159
2.2. Fitness Function and Selection
The fitness function generates a fitness value of each chromosome. The value
indicates how suitable for solving a task scheduling problem a chromosome is. The
fitness function for general task scheduling problems is based on execution time of
tasks. But, task scheduling problems in Cloud computing are different from general task
scheduling problems because computing services in Cloud computing are offered
through a SLA between cloud users and providers. The minimization of execution time
is a goal of task scheduling in Cloud computing, but the main goal is to improve user
satisfaction and increase providers profit. Therefore, the fitness function in our genetic
algorithm is defined by these elements.
The fitness function generates a fitness value of each chromosome. The value indicates
how suitable for solving a task scheduling problem a chromosome is. The fitness function for
general task scheduling problems is based on execution time of tasks. But, task scheduling
problems in Cloud computing are different from general task scheduling problems because
computing services in Cloud computing are offered through a SLA between cloud users and
providers. The minimization of execution time is a goal of task scheduling in Cloud
computing, but the main goal is to improve user satisfaction and increase provider's profit.
Therefore, the fitness function in our genetic algorithm is defined by these elements.
User satisfaction can be guaranteed with the QoS designated in a SLA document. In this
paper, response time and processing cost, which are typical QoS attributes in a SLA [2], are
used to define user satisfaction. In our scheduling algorithm, the roulette wheel selection
[11], a stochastic selection method, is used to create the basis of the next generation.
We assume that there is a roulette wheel, size of which is the sum of fitness values in a
population.
2.3. Crossover and Mutation
The crossover operation is a simple mechanism to swap one part of a chromosome
for that of another chromosome. In this paper, the two-point crossover [12] is used for
transferring genetic material of parent to children.
The mutation operation is to expand the search space by changing one part of a
chromosome. In the beginning of the genetic algorithm, the quality of generated
chromosomes is not particularly good. As time goes by, the quality of chromosomes
becomes more improved. There is the potential for improvement in quality by the
mutation operation in the early part of the genetic algorithm. But, it is hard to improve
the quality of chromosomes by the mutation operation after the quality attains a certain
standard. In this case, the mutation operation increases execution time of the genetic
algorithm. We therefore the non-uniform mutation [13], which gradually decreases the
mutation rate by the number of reproductions.
2.4. Restart & Stop Condition
Figure 2 shows the restart operation. If the best fitness value of the current
population does not reach the minimum fitness threshold, chromosomes with high
fitness values in the current population fill the half of the next population. And, the rest
of the next population is filled with chromosomes generated randomly. This restart
operation helps the GA scheduling function to prevent local optimum and explore
various search places.
International Journal of Control and Automation
Vol. 5, No. 4, December, 2012
160
Figure 2. Pseudocode of Restart Operation
And, the stop condition of the GA scheduling function is as Eq. 1 where bfc is the
best fitness value in the current population, bfp is the best fitness value in the past
population, θ is a boundary variation, and ft is the minimum fitness threshold.
)(, ctpc bffbfbf
(1)
If, the above stop condition is met, the GA scheduling function stops and outputs a
chromosome with the best fitness value, i.e., the best task schedule. Otherwise, the GA
scheduling function continues restarting or reproducing a population.
3. Simulation Results
Figure 3. Simulation Results
In this section, we simulated the GA-based Task Scheduling Model (GATSM) on the
discrete event system modeling and simulation environment [14] and conducted various
experiments to compare performances of GATSM with those of the Round Robin Task
Restart () //Restart operation
Input: a population po, a set of n tasks T, and a set of m
processors P
Output: new population po
{
1. find the chromosome bch with the best fitness value in po;
2. if (bch.fitness < ft ) { // ft is the minimum fitness threshold
3. sort po by fitness value in descending order;
4. for (i=1; i|po|/2; i++) {
5. copy po[i] to po[i]; }
6. for (j=(|po|/2)+1; j|po|; j++) {
7. po[j] = encoding(T,P); }
}
8. else {
Selection(po); }
}
International Journal of Control and Automation
Vol. 5, No. 4, December, 2012
161
Scheduling Model (RRTSM) [7], the Load Index-based Task Scheduling Model
(LITSM) [8], and the Activity Based Costing-based Task Scheduling Model (ABCTSM)
[9]. The RRTSM allocates takes to virtual machines in sequence regardless of
information on tasks or virtual machines. The LITSM uses the load index of virtual
machines for task scheduling. In the LITSM, a scheduler allocates tasks to lightly
loaded virtual machines. And, the ABCTSM uses the cost required to process tasks on
each virtual machine and allocates tasks to virtual machines as per the cost.
Performance comparisons are based on throughput, response time, virtual machine
utilization, processing cost, and user satisfaction. In experiments, 12 virtual machines
with different computing abilities were simulated. A total of 2000 tasks were generated
from cloud users each simulation. A cloud user generates a task with requirements
including time and budgetary constraints. We classified tasks into four types as shown
in Figure 1. And Figure 3 shows brief results of simulation.
4. Conclusion
In this paper, we presented scheduling problems in Cloud computing and proposed a
task scheduling model to solve the scheduling problems. In the proposed task
scheduling model, the task scheduler calls the GA scheduling function to make task
schedules based on information of tasks and virtual machines. The GA scheduling
function creates a population, a set of task schedules, and evaluates the population by
using the fitness function considering user satisfaction and virtual machine availability.
The function iterates reproducing populations to output the best task schedule. The
restart operation is also applied to the GA scheduling function for the improvement in
quality of task schedules. For performance evaluation, we simulated the proposed task
scheduling model and conducted diverse experiments. Empirical results prove that the
proposed task scheduling model outperforms existing task scheduling models, which
are the round-robin task scheduling model [7], the load index-based task scheduling
model [8], and the activity based costing based task scheduling model [9].
Acknowledgements
This research was supported by Basic Science Research Program through the
National Research Foundation of Korea (NRF) funded by the Ministry of Education,
Science and Technology (2012002751).
References
[1] B. Hayes, Cloud computing, Communications of the ACM, vol. 51, no. 7, (2008), pp. 9-11.
[2] P. Patel, A. Ranabahu and A. Sheth, Service level agreement in cloud computing,
http://knoesis.wright.edu/library/download/OOPSLA_cloud_wsla_v3.pdf, (2009).
[3] R. Buyya, C. S. Yeo, S. Venugopal, J. Broberg and I. Brandic, "Cloud computing and emerging IT
platforms: vision, hype, and reality for delivering computing as the 5th utility", Future Generation
Computer Systems, vol. 25, issue 6, (2009), pp. 599-616.
[4] W. Sun, Y. Zhang and Y. Inoguchi, "Dynamic task flow scheduling for heterogeneous distributed
computing: algorithm and strategy", IEICE Trans. on Inf. & Sys., vol. E90-D, no. 4, (2007), pp. 736-
744.
[5] I. K. Savvas and M. T. Kechadi, "Dynamic task scheduling in computing cluster environments", Proc.
3rd Int'l. Workshop on Parallel and Distributed Computing, (2004), pp. 372-379.
[6] R. Moreno, "Job scheduling and resource management techniques in dynamic grid environments ",
Lecture Notes in Computer Science, vol. 2970, (2004), pp. 25-32.
[7] K. Li, "Job scheduling and processor allocation for grid computing on metacomputers", Journal of
Parallel and Distributed Computing, vol. 65, no. 11, (2005), pp. 1406-1418.
International Journal of Control and Automation
Vol. 5, No. 4, December, 2012
162
[8] G. Kakarontzas and I. K. Savvas, "Agent-based resource discovery and selection for dynamic grids",
Proc. 15th IEEE Int'l Workshops on Enabling Technologies: Infrastructure for Collaborative
Enterprises, (2006), pp. 195-200.
[9] Q. Cao, Z. B. Wei and W. M. Gong, "An optimized algorithm for task scheduling based on activity
based costing in cloud computing", Proc. 3rd Int'l Conf. on Bioinformatics and Biomedical
Engineering, (2009), pp. 1-3.
[10] E. S. H. Hou, R. Hong and N. Ansari, "Efficient multi processor scheduling based on genetic
algorithms", Proc. IEEE Conf. on Industrial Electronics Society, vol. 2, (1990), pp. 1239-1243.
[11]G. Syswerda, "Uniform crossover in genetic algorithms", Proc. 3rd Int'l Conf. on Genetic Algorithms,
(1989), pp. 2-9.
[12] A. Neubauer, "Adaptive non-uniform mutation for genetic algorithms", Lecture Notes in Computer
Science, vol. 1226, (1997), pp. 24-34.
[13] B. P. Zeigler, H. S. Sarjoughian, S. W. Park, J. S. Lee, Y. K. Cho and J. J. Nutaro, "DEVS modeling
and simulation: a new layer of middleware", Proc. of 3rd Annual Int'l Workshop on Active
Middleware Services, (2001), pp. 22-31.
Authors
Sung Ho Jang
He received the Ph.D degree at Information Engineering from Inha
University in 2011. His research interests include Cloud computing and
Ubiquitous computing
Tae Young Kim
He received the MS degree at Computer Science and Information
Engineering from Inha University in 2009. His research interests include
Cloud computing, software modeling & simulation.
Jae Kwon Kim
He received the BS degree at Computer Science and Information
Engineering from Gachon University Medical and Science in 2011. His
research interests include Cloud computing and Ubiquitous computing
Jong Sik Lee
He is a professor in the School of Computer Science and Information
Technology. He received the Ph.D degree at Electronics and Computer
Engineering from University of Arizona in 2001. Current research
interests include Cloud computing, software modeling and
simulation.
... The evolution of metaheuristic algorithms, inspired by their applicability to both continuous and discrete problems, has led to promising approaches for the TSFC problem [13,14,15]. Early implementations of Genetic Algorithms (GA) and Particle Swarm Optimizers (PSO) demonstrated potential but were limited in scope [16,17,18,19,20,21,3,22]. ...
... This step of the algorithm mimics the good memory of marine predators that helps them remember areas of abundant food. After updating solutions using Equation 18, the fitness of solutions in the P rey matrix is evaluated to further update the Elite matrix. To mimic memory-saving, the fitness of each solution in the current iteration is compared to its respective correspondent in the previous iteration. ...
Preprint
Full-text available
Cloud and fog computing architectures are essential for decentralizing computational tasks, enabling efficient processing across user demands and extensive IoT networks. As computational loads grow in complexity and scale, effective task scheduling across these architectures becomes critical for optimizing resource usage and reducing latency. This paper introduces the Marine Predators Algorithm (MPA), a novel nature-inspired optimization approach to address the task scheduling challenge within cloud and fog environments. Due to the complexity and high-dimensionality of the scheduling problem, MPA was tested on both synthetic and real-world workloads, including HPC2N, NASA iPSC, and GOCJ datasets. Simulation results indicate that MPA outperforms several existing algorithms, achieving significant reductions in average makespan times and enhancing the Degree of Imbalancing, thereby ensuring more efficient resource distribution. These findings underscore MPA’s effectiveness in addressing the demanding task scheduling requirements of decentralized computing environments, positioning it as a promising solution for improved performance and resource utilization in cloud and fog infrastructures.
... It is a relatively new kind of computing platform that has become prevalent as a result of Internet technologies and the development of virtualization [1]. The term "cloud computing" refers to a certain kind of distributed system that is made up of a network of virtualized computers that are associated with one another and whose resources are allocated on demand. ...
... The present section presents existing research in task scheduling, cloud computing, fog computing, and metaheuristic optimization techniques. S. H. Jang [1] did a study on genetic algorithms. They focused on task scheduling in cloud computing. ...
Article
Full-text available
These days, the most popular kind of algorithm being utilized is called a meta-heuristic algorithm. Because the search space in such an algorithm can only be constrained by the best answer, the resulting searching domain is poor, which in turn leads to a long searching process. The reason for this study is to provide a comparative examination of a meta-heuristic optimization approach that may be used to address difficulties with task scheduling. When using the recently created and effective swarm intelligence algorithms, determining the solution for the Optimized task scheduling issue in Fog Environment is a tough challenge. An overwhelming number of challenges need to be tackled, including mixed decision variables; diversified restrictions; inherent mistakes; competing aims; and various locally optimum solutions. The behavior of various meta-the equilibrium optimizer, is investigated in this work. In earlier research activities, IMOMVO was developed as a solution to address the shortcomings that were discovered in the original MVO as well as its most recent improved version, MVP. This category of approaches is capable of resolving the issue of the avg positioning by improving equations for updating AP based on the best & second-best solutions currently available. The creators of IMOMVO employed many datasets scenarios with various jobs and virtual machines (Vms) to assess the capabilities of the technique while doing an evaluation. The findings of the IMOMVO approach have been validated with the use of standard evaluation criteria, including Vms processing power, task execution time, and throughput. During the task scheduling process, IMOMVO got better outcomes according to the assessment metrics than other methods that are considered to be state-of-the-art.
... Genetic Algorithm (GA) (Jang et al., 2012;Zhou et al., 2020) is a meta-heuristic algorithm that iterates over the feasible solutions by simulating the genetic and mutation processes in biological evolution, where in each generation the better-adapted individuals are selected and their genes are inherited by the next generation through crossover and mutation so that the optimized solution evolves gradually. (Al-Maamari and Omara, 2015;Pradhan et al., 2022;Kumar and Sharma, 2020) is a search algorithm based on group collaboration, which simulates the behavior of a flock of birds foraging for food. ...
Article
Full-text available
The huge energy consumption of data centers in cloud computing leads to increased operating costs and high carbon emissions to the environment. Deep Reinforcement Learning (DRL) technology combines of deep learning and reinforcement learning, which has an obvious advantage in solving complex task scheduling prob- lems. Deep Q Network(DQN)-based task scheduling has been employed for objective optimization. However, training the DQN algorithm may result in value overestimation, which can negatively impact the learning effectiveness. The replay buffer technique, while increasing sample utilization, does not distinguish between sample importance, resulting in limited utilization of valuable samples. This study proposes an enhanced task scheduling algorithm based on the DQN framework, which utilizes a more optimized Dueling-network architecture as well as Double DQN strategy to alleviate the overestimation bias and address the shortcomings of DQN. It also incorporates a prioritized experience replay technique to achieve importance sampling of experience data, which overcomes the problem of low utilization due to uniform sampling from replay memory. Based on these improved techniques, we developed an energy-efficient task scheduling algorithm called EETS (Energy-Efficient Task Scheduling). This algorithm automatically learns the optimal scheduling policy from historical data while interacting with the environment. Experimental results demonstrate that EETS exhibits faster convergence rates and higher rewards compared to both DQN and DDQN. In scheduling performance, EETS outperforms other baseline algorithms in key metrics, including energy consumption, average task response time, and average machine working time. Particularly, it has a significant advantage when handling large batches of tasks.
... ……………… (2) The study of Tae Young Kim. et al. [1] on Task Scheduling was carried out in 2012. Their research led to the development of a scheduling mechanism based on Genetic Algorithms, prioritizing user satisfaction and the availability of virtual machines. ...
Article
Full-text available
Cloud computing technology gives users more scalability and flexibility by utilizing dynamic resources. Task scheduling and resource allocation are the two main issues in a cloud environment that directly affect system throughput and user satisfaction. The two main variables influencing the cloud system's performance are the amount of time needed to complete tasks and the computational cost. A multi-objective approach to task scheduling and resource allocation is presented in this work.In order to efficiently distribute computing resources and schedule tasks while minimizing time and cost objectives, the Enhanced Honey Badger algorithm (EHBA) is employed. Next, the algorithm is combined with LSTM to further optimize it.Time-to-execute, Cost-to-compute, Task-to-resource utilization, and Time-to-respond are some of the metrics used to evaluate how well the suggested EHBA method (in conjunction with LSTM) performs for effective task scheduling and resource allocation https://doi.
... • genetic algorithm [14]- [16], • particle swarm optimization [17]- [20], • ant colony optimization algorithms [21], [22], • multi-processors-based algorithms [23], • fuzzy algorithms [24], [25], • cost-based algorithms [26]. Further, we implement only a few of these approaches. ...
Article
Full-text available
We consider the real-life problem of planning tasks for teams in a corporation, in conditions of some restrictions. The problem takes into account various constraints, such as for instance flexible working hours, common meeting periods, time set aside for self-learning, lunchtimes and periodic performance of tasks. Additionally, only a part of the team may participate in meetings, and each team member may have their own periodic tasks such as self-development. We propose an algorithm that is an extension of the algorithm dedicated for scheduling on parallel unrelated processors with the makespan criterion. Our approach assumes that each task can be defined by a subset of employees or an entire team. However, each worker is of a different efficiency, so task completion times may differ. Moreover, the tasks are prioritized. The problem is NP-hard. Numerical experiments cover benchmarks with 10 instances of 100 tasks assigned to a 5-person team. For all instances, various algorithms such as branch-and-bound, genetic and tabu search have been tested.
... In recent years, utilizing meta-heuristic methods to solve intricate optimization problems in task scheduling has also proved efficient and stable by comparing them with traditional scheduling methods. Various metaheuristic algorithms and their variants have been used to solve scheduling problems in the field of cloud computing, mainly including Particle Swarm Optimization Algorithm (PSO) , Genetic Algorithm (GA) (Jang et al. 2012), Ant Colony Optimization Algorithm (ACO) (Sun et al. 2013), Gray Wolf Optimization (GWO) (Bacanin et al. 2019), Whale Optimization Algorithm (WOA) (Sreenu and Sreelatha 2019), Taboo Search (TS), Simulated Annealing Algorithm (SAA) and Firefly Optimization Algorithm (FA) (Saleh et al. 2019). These optimization algorithms rely on models of the evolutionary processes of biological populations. ...
Preprint
Full-text available
With the popularity of cloud computing, effective task scheduling has become the key to optimizing resource allocation, reducing operation costs, and enhancing the user experience. The complexity and dynamics of cloud computing environments require task scheduling algorithms that can flexibly respond to multiple computing demands and changing resource states. To this end, this study proposes an improved RTH algorithm, the ERTH algorithm, which aims to improve the efficiency and effectiveness of task scheduling in cloud computing environments. Evaluations in the CEC benchmark test sets show that the ERTH algorithm outperforms the traditional PSO and GWO in several performance metrics and outperforms the emerging GWCA and CSA. This result signifies a significant advancement of the ERTH algorithm in intelligent optimization. Further, we apply the ERTH algorithm to a real cloud computing environment and conduct a comparison with the original algorithm RTH, PSO, ACO, WOA, and HLBO. When dealing with cloud computing task scheduling problems, the ERTH algorithm demonstrates better task completion time, resource utilization, and system load balancing performance. Especially in high-load and complex task scenarios, the stability and scalability of the ERTH algorithm perform exceptionally well. This study not only reveals the powerful potential of the ERTH algorithm in cloud computing task scheduling but also brings new perspectives and solutions for cloud service providers in resource allocation and task scheduling strategies. The proposal and validation of the ERTH algorithm are of great significance in promoting the application of intelligent optimization algorithms in cloud computing.
Article
Full-text available
This study compares the effectiveness of three optimization algorithms-Grey Wolf Optimizer (GWO), Flower Pollination Algorithm (FPA), and Particle Swarm Optimization (PSO)-for task scheduling in a cloud computing environment. The simulation results reveal that GWO consistently outperforms both FPA and PSO in terms of execution time and cost efficiency. Specifically, GWO delivers the fastest execution times, starting from 122.2 milliseconds for 200 tasks and increasing to 162.6 milliseconds for 1000 tasks. It also demonstrates superior cost efficiency, with costs beginning at Rs 123.2 for 200 tasks and rising to Rs 152.6 for 1000 tasks. In comparison, FPA and PSO show higher execution times and costs, particularly for larger task sizes. These results highlight GWO's effectiveness in managing task scheduling efficiently and economically, making it a preferable choice for large-scale and cost-sensitive cloud computing applications.
Book
Full-text available
Cloud computing that provides cheap and pay-as-you-go computing resources is rapidly gaining momentum as an alternative to traditional IT Infrastructure. As more and more consumers delegate their tasks to cloud providers, Service Level Agreements(SLA) between consumers and providers emerge as a key aspect. Due to the dynamic nature of the cloud, continuous monitoring on Quality of Service (QoS) attributes is necessary to enforce SLAs. Also numerous other factors such as trust (on the cloud provider) come into consideration, particularly for enterprise customers that may outsource its critical data. This complex nature of the cloud landscape warrants a sophisticated means of managing SLAs. This paper proposes a mechanism for managing SLAs in a cloud computing environment using the Web Service Level Agreement(WSLA) framework, developed for SLA monitoring and SLA enforcement in a Service Oriented Architecture (SOA). We use the third party support feature of WSLA to delegate monitoring and enforcement tasks to other entities in order to solve the trust issues. We also present a real world use case to validate our proposal.
Conference Paper
Full-text available
In this study, a cluster-computing environment is employed as a computational platform. In order to increase the efficiency of the system, a dynamic task scheduling algorithm is proposed, which balances the load among the nodes of the cluster. The technique is dynamic, nonpreemptive, adaptive, and it uses a mixed centralised and decentralised policies. Based on the divide and conquer principle, the algorithm models the cluster as hyper-grids and then balances the load among them. Recursively, the hyper-grids of dimension k are divided into grids of dimensions k - 1, until the dimension is 1. Then, all the nodes of the cluster are almost equally loaded. The optimum dimension of the hyper-grid is chosen in order to achieve the best performance. The simulation results show the effective use of the algorithm. In addition, we determined the critical points (lower bounds) in which the algorithm can to be triggered.
Conference Paper
Full-text available
The massive amount of resources on computational grids raises the question of efficient resource discovery and selection. In this paper we present an agent-based approach to these two phases of a grid scheduling process. The approach is based on client agents which act on behalf of grid users, and search for resources in a network of resource representatives that are registries of resource characteristics. After potentially suitable resources are discovered, client agents carry out negotiations directly with agents representing local resource management systems. This selection process first determines the willing resources with a contract-net protocol and then uses current load information and a more accurate post-discovery search to determine the optimum set of resources for the task in hand.
Conference Paper
Full-text available
In this paper, we analyze the problem of grid resource broker- ing in the presence of economic information about the price of resources. We examine in detail the main tasks that a resource broker has to carry out in this particular context, like resource discovery and selection, job scheduling, job monitoring and migration, etc. Then, we propose an ex- tension of the grid resource information service schema to deal with this kind of economic information, and we evaluate different optimization criteria for job scheduling and migration, combining both performance and economic information. The experimental application benchmark has been taken from the finance field, in particular a Monte Carlo simulation for pricing European financial options.
Conference Paper
An efficient method, based on genetic algorithms, for solving the multiprocessor scheduling problem is proposed. The representation of the search node is based on the schedule of the tasks in each individual processor. The genetic operator is based on the precedence relations between the tasks in the task graph. The genetic algorithm is applied to the problem of scheduling robot inverse dynamics computations
Conference Paper
In cloud computing, traditional way for task scheduling cannot measure the cost of cloud resources accurately by reason that each of the tasks on cloud systems is totally different between each other. There may be no relationship between the overhead application base and the way that different tasks cause overhead costs of resources in cloud systems. The traditional way for task scheduling cannot meet the cloud market well enough. This paper introduces an optimized algorithm for task scheduling based on ABC (activity based costing) in cloud computing and its implementation. Compared with the traditional methods of task scheduling, a new method with an optimized algorithm based on ABC algorithm was proposed in this paper.
Article
With the significant advances in Information and Communications Technology (ICT) over the last half century, there is an increasingly perceived vision that computing will one day be the 5th utility (after water, electricity, gas, and telephony). This computing utility, like all other four existing utilities, will provide the basic level of computing service that is considered essential to meet the everyday needs of the general community. To deliver this vision, a number of computing paradigms have been proposed, of which the latest one is known as Cloud computing. Hence, in this paper, we define Cloud computing and provide the architecture for creating Clouds with market-oriented resource allocation by leveraging technologies such as Virtual Machines (VMs). We also provide insights on market-based resource management strategies that encompass both customer-driven service management and computational risk management to sustain Service Level Agreement (SLA)-oriented resource allocation. In addition, we reveal our early thoughts on interconnecting Clouds for dynamically creating global Cloud exchanges and markets. Then, we present some representative Cloud platforms, especially those developed in industries, along with our current work towards realizing market-oriented resource allocation of Clouds as realized in Aneka enterprise Cloud technology. Furthermore, we highlight the difference between High Performance Computing (HPC) workload and Internet-based services workload. We also describe a meta-negotiation infrastructure to establish global Cloud exchanges and markets, and illustrate a case study of harnessing ‘Storage Clouds’ for high performance content delivery. Finally, we conclude with the need for convergence of competing IT paradigms to deliver our 21st century vision.
Conference Paper
A theoretical analysis of Michalewicz' non-uniform mutation operator is presented and a novel variant — the adaptive non-uniform mutation operator — is proposed. The non-uniform mutation operator was developed by Michalewicz' for his modified variant of genetic algorithms modGA to tackle numerical parameter optimization problems. As is shown by mathematical analysis, this mutation operator prefers parameter values in the center of the corresponding feasible region. This leads to problems if the optimum is situated near the feasible region's boundaries. In order to avoid this undesirable tendency, the adaptive non-uniform mutation operator is proposed, the development of which rests on the mathematical analysis. Experimental results for a standard numerical parameter optimization problem are given that illustrate the superiority and effectiveness of this novel mutation operator for genetic algorithms.
Conference Paper
This paper discusses a concept of DEVS middleware that would provide a set of services for constructing discrete event models and executing them in simulation or in real-time. Although generic middleware such as CORBA and MPI supports simulation and simulation-oriented middleware such as High Level Architecture Runtime Infrastructure exists, there are many issues that either not addressed at all or have been inadequately addressed in existing systems. DEVS Middleware would improve modeling and simulations due to the beneficial formal properties of DEVS, enhance interoperability among components adhering to the DEVS protocol, and reduce programming complexity by hiding lower level Simulation and execution details. The DEVS Middleware concept includes components for model construction and for mapping models onto simulators or real-time execution engines to support the novel concept of "distributed programming by modeling." This paper discusses several sub-layers that have been identified within the overall DEVS middleware concept, existing work that can be brought to bear on developing these layers and the integration needed to create a widely accepted standard.