Qo s aware scientific application scheduling algorithm in cloud environment

Computer Engineering and Intelligent Systems
ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online)
Vol.4, No.12, 2013

www.iiste.org

QoS-aware Scientific Application Scheduling Algorithm in Cloud
Environment
Ayda Mazandarani (Corresponding author)
Mirdamad University of Gorgan, Gorgan, Iran
E-mail: ayda.mazandarani@gmail.com
Hossein Momeni
Gorgan University of Agricultural sciences and Natural Resources, Gorgan, Iran
E-mail: momeni@Gau.ac.ir
Abstract
Many complex scientific applications are modeled in the form of workflows to carry out large-scale experiments.
Because of complexity of scientific processes, scientific workflows need intensive computation and data
requirements. Clouds make opportunity for scientific that need high performance computing infrastructure. So
scientific can run their application on cloud by their desired QoS. We propose an algorithm that able scientific to
select execute plan based on their preference QoS, like time and cost. Proposed algorithm ranks the tasks in
workflow and then use UPFF function for select accurate resource, based on user’s QoS. We compared our
proposed algorithm with the same work by several scenarios and results show proposed algorithm has better
efficiency.
Keywords Scientific application, Workflow scheduling, Cloud computing
1.

Introduction

The field of distributed and parallel computing has seen technologies rapidly grow from desktop computing,
through grid computing, and now to cloud computing. All these technologies focus on delivering computing
power to a large number of end-users in a reliable, efficient and scalable manner. More and less, the trend has
been to deliver the computing power as a utility, much like how water and electricity is delivered to households
these days.
Cloud computing is a kind of parallel and distributed computing systems that delivers infrastructure, platform
and software as a service, which are made available as services in a pay-as-you-go model to consumer. These
services are referred to as infrastructure as a service (IaaS), platform as a service (PaaS), software as a service
(SaaS). In [3] Buyya et al. define a cloud as a “type of parallel and distributed system consisting of a collection
of interconnected and virtualized computers that are dynamically provisioned and presented as a one or more
unified computing resources based on service-level agreements”. Clouds try to make opportunity to the users all
over the world to be able access the services on demand, according to their desired quality of service
requirements. So it offers lots of benefits for companies by decreasing management and maintenance costs from
leasing IT infrastructure from cloud providers.
Many scientific applications in the field of astronomy, gravitational-physics, computational biology, climate
modeling, and life-sciences have used workflow technology to carry out large-scale experiments. Scientific
applications are typically modeled as workflows that consist of tasks, data, control sequences and data
dependencies [7].
Because of complexity of scientific process, these applications should be usually run on the large and distributed
computing environments like cloud environment. Clouds present a chance for scientists whom need highperformance computing infrastructure for their experiments[9]. Most of the time, applications are represented as
a scientific workflows that can manage many activities and work with lots of data. Scientific already using cloud
computing that schedule these workflows onto distributed cloud resources for optimizing various objectives:
minimize total makespan of the workflow, minimize cost and usage of network bandwidth, minimize cost of
computation and storage, meet the deadline of application, and combination of objectives.
A data intensive computing environment consists of applications that produce, manipulate, or analyze data in the
range of hundreds of megabytes (MB) to petabytes (PB) and beyond. A data intensive application workflow has

21

Vol.4, No.12, 2013

www.iiste.org

comparatively higher data workloads to manage than its computational load. In the words, the requirements of
resource interconnection bandwidth for transferring data outweigh the computational requirements for
processing tasks. As a consequence, demands more time to transfer and store data as compared to execution time
for tasks in the workflow. It is common to characterize the distinction between data intensive and compute
intensive by defining a threshold for the computation to communication ratio (CCR). Application with lower
values of this ratio is distinctly data intensive in nature [8].
The rest of paper is organized as follows: section 2 presents related work. In section 3, we describe the taskresource scheduling problem. In section 4, we present our scheduling algorithm. Section 5 presents an
experimental evaluation of the performance our protocol. Section 6 concludes the paper and discusses some
future work.
2.

Related work

Workflow applications are commonly represented as a directed acyclic graph. The mapping of jobs to the
compute-resources is an NP-complete problem in the general form. A survey of field of scientific workflows and
techniques for managing and scheduling them are represented in [7]. [8] is a PHD theses that represented several
scientific workflow scheduling algorithms in grid and cloud environments.
In [28] proposed an optimized scheduling algorithm to achieve the optimization or sub-optimization for cloud
scheduling. In this algorithm an Improved Genetic Algorithm (IGA) is used for the automated scheduling policy.
It is used to increase the utilization rate of resources and speed.
In [27] proposed an improved cost-based scheduling algorithm for making efficient mapping of tasks to available
resources in cloud. This scheduling algorithm measures both resource cost and computation performance, it also
Improves the computation/communication ratio.
In [29] proposed an SHEFT workflow scheduling algorithm to schedule a workflow elastically on a Cloud
computing environment. The experimental results show that SHEFT not only outperforms several representative
workflow scheduling algorithms in optimizing workflow execution time, but also enables resources to scale
elastically at runtime.
In [18] presented a particle swarm optimization (PSO) based heuristic to schedule applications to cloud resources
that takes into account both computation cost and data transmission cost. It is used for workflow application by
varying its computation and communication costs. The experimental results show s that PSO can achieve cost
savings and good distribution of workload onto resources.
In [19] proposed a market-oriented hierarchical scheduling strategy which consists of a service-level scheduling
and a task-level scheduling. The service-level scheduling deals with the Task-to-Service assignment and the
task-level scheduling deals with the optimization of the Task-to-VM assignment in local cloud data centers.
In [20] worked on multiple workflows and multiple QoS. They has a strategy implemented for multiple
workflow management system with multiple QoS. The scheduling access rate is increased by using this strategy.
This strategy minimizes the make span and cost of workflows for cloud computing platform.
In [21] presented the HEFT algorithm. This algorithm first calculates average execution time for each task and
average communication time between resources of two successive tasks. Then tasks in the workflow are ordered
(non-increasing) on a rank function. The task with higher rank value is given higher priority. In the resource
selection phase tasks are scheduled in the order of their priorities and each task is assigned to the resource that
can complete the task at the earliest time.
In [24], data resource replication and parallel retrieving from several location are used to reach efficient
scheduling plan. ESMH algorithms based on retrieving from several data resources are presented that mapped
tasks on resources, according to data retrieving time and task compute time.
3.

Problem definition

A scheduling system model consists of an application, a target cloud computing environment, and a performance
criteria for scheduling. We denote an application workflow as a directed acyclic graph (DAG) represented by
G=(V,E), where V={T1, … , Tn} is the set of tasks, and E represents the data dependencies between these tasks.
The number of tasks are considered n. G is a n*n matrix of Directed acyclic graph that gained by scientific
application. if the amount of Gi,j is one, represent task Ti and task Tj are dependable and Ti is the parent of Tj, else
if the amount of Gi,j is zero, shows they aren’t dependable. if tasks were dependable, the data of parent task Ti
should be transmitted to a child task Tj. a task without any parent is called an entry task and a task without any

22

Vol.4, No.12, 2013

www.iiste.org

child is called an exit task. The size of output data of tasks are given in array D as Di represent size of output
data task Ti.
We assumed that we want to use some of resources in cloud for scheduling. We have a set of resources R =
{1,… , m}. here, we consider the number of resources is m. Each resource has its feature like cost of running
tasks or Cost_exe(Ri), cost of incoming data by resources or Cost_in(Ri), cost of sending data from resources or
Cost_out(Ri), resource availability or Availability(Ri) and resource reliability or Reliability(Ri).
The estimated times for compute each task on each resource are given. W is the n*m matrix that represent the
estimated times for execute each task on each resource. Wi,j shows the estimated time for execute task Ti on
resource Rj .
The objective function of our workflow scheduling problem is to determine the schedule plan to assign tasks of a
given scientific application to target cloud resources such that it’s been done in user’s desired time and cost.
4.

Proposed approach

We present our proposed approach as a scheduling algorithm. The key idea of our approach is based on HEFT
algorithm [21]. We named the proposed approach as a QoS-aware Scientific Application Scheduling Algorithm
(QSASA). In QSASA we try to preserve benefits of HEFT and also apply the data-intensive scheduling features
in cloud environment. QSASA considers varied aspects of scientific workflow scheduling. Aspects like tasks
dependency, tasks data size, compute time of tasks, data transfer time from parent task to child, workflow
makespan, resources bandwidth, resources cost for compute, resources cost for input or output data, availability
and reliability parameters.
Before presenting the proposed approach in section 4-1, we formulate the parameters and attributes that are used
in proposed protocol. And in the section 4-2 we present our protocol.
4.1 Problem formulation

In this section, we present the metrics of comparison, the experiment setup and the results.In
section 3 we explained the problem and the parameters and information that gained by given scientific
application and target cloud resources. Also there are several attributes that we may use in the proposed protocol.
In this section, we formulate the attribute that may used in proposed protocol.
As we said, W is the n*m matrix that represent the estimated times for execute each task on each resource and
Wi,j shows the estimated time for execute task Ti on resource Rj. before scheduling, the tasks are labeled with the
average execution times. The average execution time of a task Ti is defined as

Target cloud resources are connected and so considering the communication time should be occurred. Resource
bandwidth or data transfer rates are stored in matrix B of size m*m. the amount of Bi,j represents resource
bandwidth between resource Ri and Rj. the communication time of the edge (i,k), which is for transferring data
from task Ti (scheduled on Rg) to task Tk (scheduled on Rh), is defined by

When both Ti and Tk are scheduled on the same resource,
becomes zero since we assume that
the intra-resource communication time is negligible when it is compared with the inter-resources communication
time. Before scheduling, average communication times are used to label the edges. The average communication
time of an the edges out from Ti, is defined by

Where

is the average transfer rate among the target cloud resources.

It is necessary to reword some attributes like EST(Ti, Rj) and EFT(Ti, Rj) in [21] that represent earliest start time
and earliest finish time task Ti on resource Rj, respectively. For the entry task Tentry,
EST(Tentry, Rj) = 0

23

(4)

Vol.4, No.12, 2013

www.iiste.org

For the rest tasks in workflow, the EST and EFT values are computed respectively, starting from the entry task,
as shown in (5) and (6), respectively. In order to compute the task Ti, all immediate predecessor tasks of Ti must

have been scheduled.
Where pred(Ti) is the set of immediate predecessor tasks of task Ti and avail[j] is the earliest time at which
resource Rj is ready for task execution. The inner max block in the EST equation returns the ready time, means
the time when all data needed by Ti has arrived at resource Rj.
ESTmin shows minimum value of EST for task on target resources and ESTmax shows maximum value of EST. To

compute average values, we used

instead of B for bandwidth and shown by

(Ti) and

(Ti).

EFC is another attribute that represent total cost for execute tasks containing run cost, input data cost and output
data cost that are calculate by (11).

Where p is the number of parent tasks of task Ti. communicatek,I is the required time for transmitting data from
parent tasks of Ti, Tk, that is mapped on resource Rl, to task Ti that mapped on resource Rj. EFCmin represents
minimum value of EFC of task T on all target resources and EFCmax shows maximum value of EFC.
Tasks are sorted by their priority. Upward ranking [21] of task Ti are calculated recursively by (12).

Where succ (Ti) is the set of immediate successors of task Ti. the rank is computed recursively by traversing the
task graph upward, starting from exit task. For exit task Texit, the upward rank is calculated by (13).
Ranku(Texit) =

exit

(13)

4.2 Proposed Algorithm
We present QoS-aware scientific application scheduling algorithm (QSASA) as a proposed approach. We aimed
to schedule scientific workflow on target cloud resources based on user’s preferences. QSASA algorithm is
shown in figure 1. and have two phases, ranking tasks and selecting resources. Average time values of execute
tasks and edge communication are computed. Then values for EST, EFT and EFC for each task on all resources
are calculated. According to calculated attributes and the parameters, we use upward ranking for ranking tasks in
workflow and use User preference Fitness Function (UPFF) for accurate resource selection. And last tasks
dispatched on resources for workflow execution.
BEGIN : QoS-aware Scientific Application Scheduling Algorithm (QSASA)
INPUT:
A matrix G represented scientific workflow graph
A array D represented data produced by tasks
A matrix B represented bandwidth between resources
A matrix W represented times for execute tasks on resources
A array Cost_exe represented cost of execute tasks on resources

24

Vol.4, No.12, 2013

www.iiste.org

A array Cost_in represented cost of input required tasks on resources
A array Cost_out represented cost of output required tasks on resources
A array Reliability represented reliability of resources
A array Availability represented availability of resources
PRE-COMPUTE:
Compute average time of execution for each task,
Compute average time of communication for each edge (task dependency),
Compute average value of earliest start time for each task,
Compute average value of earliest finish time for each task,

(Ti)
(Ti)

Ranking tasks by upward ranking in [21] as Rank_upward list
While all the tasks aren’t scheduled
Start for all resources in R
Calculate EFT for task T on all resources in R
Calculate Costs for all resources in R that their EFT was computed (EFC)
Determinate ESTmin (T), ESTmax (T), EFCmin (T), EFCmax (T)
Use UPFF function for select accurate resource
End
Map task T on the resource Rthat have minimum value of UPFF function
End
Distributed tasks on resources
END : QoS-aware Scientific Application Scheduling Algorithm (QSASA)
Figure 1.QoS-aware Sientific Application Scheduling algorithm (QSASA)
As we said, we used upward ranking [21] to make a priority list as Rank_upward list. UPFF function is used to
find resources for tasks in Rank_upward list, respectively. based on the user’s preference, we use different values
for weight of time weightt and weight of cost weightc. this two weight are in the range of [0, 1] and the sum of
them is 1. For example if weightc be equal to 0.7, represents 70% user is concerned with cost and 30% is
concerned with time. So the UPFF is a two variable function. As the time and cost aren’t the same kind, we
normalize these values in the range of [0, 1]. UPFF is defined by (14).

UPFF represents fitness of resource Rj for task Ti.the resourcewith minimum UPFF is selected and the task is
mapped on it. the weightt and weightc make a chance to scientific for their desired QoS requirements.
5.
Experiment evaluation
In this section, we present the metrics of comparison, the experiment setup and the results. In this evaluation, the
value for weight of cost 0.7 and the value for weight time 0.3 are considered. We assume user is most interested
in minimizing cost.
5.1 Performance metrics
As a measure of performance, we used time and cost for complete execution of application as a metrics. We
computed the total cost of execution of a workflow using two approaches: QSASA protocol and HEFT
algorithm.
5.2Illustrative example

25

Vol.4, No.12, 2013

www.iiste.org

Figure 4(a) depicts a workflow structure with ten tasks, which are represented as nodes. The dependencies
between tasks are represented as arrows. Each task generates output data after it has completed. These data are
used by the task’s children, if any. Figure 4(b) depicts three resources interconnected with varying bandwidth
and having its own execution, input and output costs. Also the estimated average time for each tasks on all
resources are given. In this example we consider that the resources are completely reliable and available.

(a) Workflow and the output data of each tasks (b) Resources and the estimated times for execute tasks

Figure 4. Illustrative example of workflow and target cloud resources
We use QSASA algorithm and HEFT algorithm on the example and the results are gained as shown in Figure 5
(a) and Figure 5(b). As you can see, QSASA algorithm has a higher efficiency than HEFT algorithm, both in
time and cost perspective.

(a) Comparison of total cost

(b)

Comparison of total time

Figura 5. Comparison between QSASA and HEFT for an illustrative example workflow and target resources
5.3Data and Implemention
We have used several matrix and array that store the values like average time of execute tasks on resources,
average communication cost between resources, data size of tasks, execute cost, input/output cost and so on.
The values for Cost_in(Ri) and Cost_out(Ri) resemble the cost of unit data transfer between resources given by
Amazon CloudFront. We randomly use that values for each iterative of experiment. In some experiments, we use
the Amazon EC2’s pricing policy for different classes of virtual machine instances. As each task has its own
data, the sum of all the data values varies according to the size of data we experiment from 64 MB to 1024 MB.

26

Vol.4, No.12, 2013

www.iiste.org

We assumed in this work, the weightc is equal to 0.7 and the weightt is equal to 0.3 and are fixed in all
experiment. So we assumed user wants more to cost saving than time saving.
We use C++ environment to conduct our simulation experiment. In addition we implement HEFT algorithm, too.
Also we have done twenty independent executions for all scenarios to gain better results.
5.4Experiment and Result
Each task in workflow has input/output data in varying sizes. We evaluate proposed protocol in different
scenarios. We plot the graphs by varying the results obtained after twenty independent executions. In almost
every execution, the x-axis parameters present number of tasks in workflow, total data size and costs range. The
y-axis parameters present performance criteria like time and cost.
5.4.1 Variation in number of workflow tasks
In this scenario, we try to compare QSASA with HEFT algorithm when the number of tasks in workflow is
varying from 10-50 tasks. We fixed the compute resource cost in the range 0.30-0.80$/hr and the communicate
cost in the range 0.14-0.38$/hr, in the sub-section 5-4-1 and 5-4-2. by varying number of tasks in workflow, we
compared SWASP and HEFT algorithm, time and cost perspective and the results depicted in Figure 6(a). and
Figure 6(b). In both figures, x-axis represents number of tasks in workflow. Based on the results, QSASA
algorithm has better performance.

(a)


(b)

Comparison of total cost

Figure 6. Comparison between QSASA and HEFT when while varying number of tasks in workflow
5.4.2 Variation in total data size of a workflow
We varied the size of total data processed by the workflow in the range 64—1024MB. By varying the data size,
we compared the variance in total time and cost of execution, for two approaches as depicted in Figure 7(a) and
Figure 7(b). Results show that QSASA has better performance. In some experiment, the QSASA’s time criteria
may be higher than HEFT, but usually the cost criteria is lower than.

(a) Comparison of total cost

(b)


Figure 7. Comparison between QSASA and HEFT when while varying number of tasks in workflow
5.4.3 Variation in resource cost
We experiment the performance of QSASA by varying the cost of computation of all target resources. This
variation is practically justifiable as different cloud service providers (e.g. Amazon) can have varying pricing
policies depending on the type and capabilities of their resources. Figure 8 depicts the change in total cost of

27

Vol.4, No.12, 2013

www.iiste.org

computation for different range of resource price. The workflow processed a total of 128MB of data and the xaxis represents range of resources cost. As QSASA consider the resource cost, Results show by varying resource
cost, it has low variance.

Figure 8. Total cost by QSASA when while varying cost of compute resources
5.4.4 Scheduling Failure variance
There are some communications between resources to execute workflow. This communication is done by
resource bandwidth. Communications may encounter delay. We want to use a approach that consider this
condition. According to the history of resources, values as reliability and availability factor are assigned to the
resources. QSASA considers these factors. In Figure 9 amount of scheduling variation are shown. X-axis shows
multiply of average values of reliability and availability.

Figure 9. Comparison of scheduling failure variance between QSASA and HEFT when while varying
availability and reliability of resources.
6.

Conclusion and future work

In this work, we presented a scientific workflow scheduling algorithm on cloud resources as called QSASA,
based on HEFT algorithm. QSASA contains two phases for ranking tasks and selecting resources. We used
QSASA algorithm to minimize the total cost of execution of application workflows on cloud environment, based
user’s preference. In evaluation we consider the user is more interested to minimize cost. We compared the
results obtained by our algorithm against HEFT algorithm. We found that QSASA achieve at least 15 percent
improvement on cost saving and at last 4 percent more time as compared to HEFT for our experiments. In
addition, QSASA considers reliability and availability factor of resources while scheduling.

28

Vol.4, No.12, 2013

www.iiste.org

As part of our future work, we would like to expand our work for independent compute resource and store
resource. Also we want to work on scheduling scientific application on BIG DATA.
References
1.
2.
3.

4.

5.

6.
7.
8.
9.

10.

11.

12.
13.
14.
15.
16.
17.
18.

19.
20.

21.
22.

Kleinrock L (2003) An Internet Vision: The Invisible Global Infrastructure. Ad-Hoc Networks vol 1, no
1, pp 3–11
Vaquero LM, Rodero-Merino l, Caceres J, Lindner M (2008) A break in the clouds: towards a cloud
definition. SIGCOMM Computer Communication Review, vol 39, pp 50–55
Buyya R, Yeo CS, Venugopal S, Broberg J, Brandic I (2009) Cloud computing and emerging IT
platforms: Vision, hype, and reality for delivering computing as the5th utility. Future Generation
Computer Systems, vol 25, no 6, pp 599–616
Duncan D, Chu X, Vecchiola C, Buyya R (2009) The Structural of the New IT Frontier: Cloud
Computing - Part Ι. CLOUDS Laboratory, Department of Computer Science and Software Engineering,
the University of Melbourne, Australia
Vecchiola C, Duncan D, Buyya R (2009) The Structural of the New IT Frontier: Market Oriented Cloud
Computing - Part Π. CLOUDS Laboratory, Department of Computer Science and Software Engineering,
the University of Melbourne, Australia
Foster I, Zhao I, Raicu I, Lu S (2008) Cloud Computing and Grid Computing 360-Degree Compared.In :
Grid Computing Environments Workshop, 2008 GCE '08, pp1-10
Pandy S, Buyya R (2009)Scheduling and Management Techniques for Data-Intensive Application
Workflows. IGI Global, USA
Pendy S (2010) Scheduling and Management of Data Intensive Application Workflows in Grid and Cloud
Computing Environments. Ph.D Thesis, University of Melbourne, Australia
Lee J, Tierney B, Johnston WE (1999) Data Intensive Distributed Computing; AMedical Application
Example. in HPCN Europe ’99: Proceedings of the 7th InternationalConference on High-Performance
Computing and Networking. London,UK:Springer-Verlag, pp 150–158
Zhao Y, Raicu I, Foster I (2008) ScientificWorkflow Systems for 21st Century, NewBottle or New
Wine?.in SERVICES ’08: Proceedings of the 2008 IEEE Congress onServices - Part I. Washington DC
USA: IEEE, pp 467–471
Buyya R, Pandey S, Christian V (2009) Cloudbus Toolkit for Market-Oriented Cloud Computing.
Proceeding of the 1st International Conference on Cloud Computing (CloudCom 2009, Springer,
Germany), Beijing, China
Juve G, Deelman E, Vahi K, Mehta G, Berriman B, Maechling P (2009) Scientific workflow applications
on Amazon EC2. in E-Science Workshops, 2009 5th IEEE International Conference pp 59-66
Esteves R (2011) A Taxonomic Analysis of Cloud Computing. 1st Doctoral Workshop in Complexity
Sciences, ISCTE-IUL/FCUL
Taylor IJ, Deelman E, Gannon DB, Shields M (2007)Workflows for e-Science: Scientific Workflows for
Grids. 1 ed. Springer
Oliveira D, Baião F, Mattoso M (2010) Towards a Taxonomy for Cloud Computing from an e-Science
Perspective. Cloud Computing: Principles, Systems and Applications, Heidelberg: Springer- Verlag
Juve G, Deelman E (2010) Scientific workflows and clouds. Crossroads v 16 (Mar.), p 14–18
Deelman E, Singh G, et al. (2008) The Cost of Doing Science on the Cloud: The Montage Example. in
SC'08 Austin, TX2008
Pandey S, Wu L, Guru S, Buyya R (2010) A Particle Swarm Optimization (PSO)-based Heuristic for
Scheduling Workflow Applications in Cloud Computing Environments. Proceedings of the 24th IEEE
International Conference on Advanced Information Networking and Applications (AINA 2010) Perth
Australia
Wu Z, Liu X, Ni Z, Yuan D, Yang Y(2011)A market-oriented hierarchical scheduling strategy in cloud
workflow systems. The Journal of Supercomputing pp 1–38
Xu M, Cui L, Haiyang W, Yanbing B(2009) A Multiple QoS Constrained Scheduling Strategy of
Multiple Workflows for Cloud Computing. in Parallel and Distributed Processing with Applications 2009
IEEE International Symposium pp. 629-634
Topcuoglu H, Hariri S, Wu M. Performance effective and low-complexity task scheduling for
heterogeneous computing IEEE Transactions on Parallel and Distributed Systems 13(3):260–274
Sakellariou R, Zhao H (2004) A Hybrid Heuristic for DAG Scheduling on Heterogeneous Systems. The
13th Heterogeneous Computing Workshop (HCW 2004) Santa Fe New Mexico USA

29

Vol.4, No.12, 2013

www.iiste.org

23. Bajaj R. Agrawal DP (2004) Improving Scheduling of Tasks in a Heterogeneous Environment. IEEE
Transactions on Parallel and Distributed Systems 15:107-118
24. Pendy S, Buya R (2010) Scheduling Data Intensive Workflow Applications based on Multi-Source
Parallel Data Retrieval in Distributed Computing Networks. the Computer Journal
25. Hoffa C, Mehta G, Freeman T, Deelman E, Keahey K, Berriman B, Good J (2008) On the Use of Cloud
Computing for Scientific Workflows. in IEEE Fourth International Conference on eScience pp 640-645
26. Lin C, Shiyong L (2011) Scheduling Scientific Workflows Elastically for Cloud Computing. in Cloud
Computing (CLOUD) 2011 IEEE International Conference on pp 746-747
27. Selvarani S, Sahasivam GS (2010) Improved Cost-Based Algorithm for Task Scheduling in Cloud
Computing. Department of Information Technology Tamilnadu Colledge of Engineering Coimbatore
India IEEE
28. Ge Y, Wei G (2010) GA-Based Task Scheduler for the Cloud Computing Systems. International
Conference on Web Information Systems and Mining
29. Bajaj R, Agrawal DP (2004) Improving Scheduling of tasks in a Heterogeneous Environment. IEEE
Transactions on Parallel and Distributed systems vol 15 pp 107-118

30

This academic article was published by The International Institute for Science,
Technology and Education (IISTE). The IISTE is a pioneer in the Open Access
Publishing service based in the U.S. and Europe. The aim of the institute is
Accelerating Global Knowledge Sharing.
More information about the publisher can be found in the IISTE’s homepage:
http://www.iiste.org
CALL FOR JOURNAL PAPERS
The IISTE is currently hosting more than 30 peer-reviewed academic journals and
collaborating with academic institutions around the world. There’s no deadline for
submission. Prospective authors of IISTE journals can find the submission
instruction on the following page: http://www.iiste.org/journals/
The IISTE
editorial team promises to the review and publish all the qualified submissions in a
fast manner. All the journals articles are available online to the readers all over the
world without financial, legal, or technical barriers other than those inseparable from
gaining access to the internet itself. Printed version of the journals is also available
upon request of readers and authors.
MORE RESOURCES
Book publication information: http://www.iiste.org/book/
Recent conferences: http://www.iiste.org/conference/
IISTE Knowledge Sharing Partners
EBSCO, Index Copernicus, Ulrich's Periodicals Directory, JournalTOCS, PKP Open
Archives Harvester, Bielefeld Academic Search Engine, Elektronische
Zeitschriftenbibliothek EZB, Open J-Gate, OCLC WorldCat, Universe Digtial
Library , NewJour, Google Scholar

Qo s aware scientific application scheduling algorithm in cloud environment

More Related Content

What's hot (18)

Similar to Qo s aware scientific application scheduling algorithm in cloud environment (20)

More from Alexander Decker (20)

Recently uploaded (20)

Qo s aware scientific application scheduling algorithm in cloud environment