International Journal of Research & Development in

Technology and Management Science –Kailash

Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

A GREEDY ALGORITHM FOR TASK SCHEDULING & RESOURCE

ALLOCATION PROBLEMS IN CLOUD COMPUTING
By
Lal Shri Vratt Singh | Research Scholar | [email protected] | Dept. of Computer Science,
Jamia Hamdard University
And
Jawed Ahmed | Assistant Professor | [email protected] | Dept. of Computer Science,
Jamia Hamdard University

ABSTRACT
Cloud computing is service based, on demand and pay per usage
computing where elastic and scalable IT functionalities are delivered
to multiple user ends using internet as a medium Distributed
computing, utility computing and virtualization play a major role to
provide an efficient base for the building of cloud computing
concept. It gives a virtual pool of computing resources which are
used by the users over the internet. After looking the benefits and
advantages of cloud computing, the market is shifting from
traditional technologies to cloud based technologies. Resource
allocation is a concern that is used in many computing fields, such as
datacenter management and operating systems. In cloud systems,
resource allocation scheme can be used as a mechanism that aims to
ensure that the applications' requirements are properly and correctly
fulfilled by the provider's infrastructure. The providers of cloud
computing services keep this most important thing in their mind that
the resources are to be used efficiently so that they may gain
maximum profit. This leads resource allocation and task scheduling
as a core and challenging issue in cloud systems.

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

This paper deals with some efficient task scheduling and resource
allocation algorithms that give the optimum result in cloud
computing environment.

KEYWORDS: Cloud computing, task scheduling, resource allocation algorithm, activity

based costing, optimized algorithm.

I. INTRODUCTION
The approach of cloud computing has become very much popular to support and operate on
demand services. It is a computing style where elastic and scalable IT resource functionalities
are delivered to the users
Figure 1: Cloud computing architecture
using internet as
services. In the cloud
environment, the
scheduling strategy
determines the best
selection of computing
resources for efficient
and the effective
execution of the tasks
after considering the
restrictions and Static /
dynamic behavior of subsequent tasks. As the cloud computing is becoming popular, the
scheduling mechanism of cloud computing is receiving a big growing attention. Basically,
scheduling is the process in which the tasks are mapped to the available resources after
looking the requirements and characteristics of the tasks. It is essential concern while dealing
with cloud environment that different task parameters need to be kept into consideration in
order to achieve efficient scheduling.

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

II. PROBLEMS WITH RESOURCE ALLOCATION AND TASK

SCHEDULING IN CLOUDE COMPUTING

A. RESOURCE ALLOCATION IN CLOUD COMPUTING

The Computational resources may be required to the user for specific time duration of a task.
For example, a large number of computer systems can be required by a scientist to simulate a
project for few hours, but it might not need, at any specific period of time. A software trainer
may want to set up a cluster of computer systems with specific software configuration that is
to be available to the trainee software professionals during the training sessions at specific
times during the week. A telecom
organization may process its
existing setup that allows the
hosting of many websites, but may
need to substitute that setup in
addition with more resources
during the period of increasing web
traffic. In other words, those
resources have to be put ready and
available right away with very

Figure 2: Inputs in the allocation of resources

advance notice [1]. The resource
allocation mechanism has been used in many areas of computing, such as grid computing,
operating systems and datacenter management. In the cloud environment, the resource
allocation system can be looked as any mechanism that ensures the correct mapping of
applications’ requirements to the providers’ infrastructure. Apart from this assurance to the
developer, the current status of the each resource should be considered by the resource
allocation mechanism in the cloud environment so that the implementation of algorithm may
allocate physical and/or virtual resources more efficiently to the cloud applications for
minimizing the operational cost. [6]

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

ALLOCATION OF VIRTUAL MACHINES (VMs):

Allocation of virtual machines can be divided into parts. The first part considers the new
requests for virtual machine provisioning and the transferring of virtual machines on hosts.
The second part concentrates on optimizing the current allocation of virtual machines. If we
talk about the complexity of the allocation part of the algorithm, it would be ‘nXm’ which
can be analyzed by calculating the number of virtual machines ‘n’ which have to be allocated
and the number of hosts ‘m’. Further the optimization of virtual machines’ current allocation
is brought out in two steps- (i) selection of the virtual machines that need to be transferred
and (ii) placement of the selected virtual machines on the hosts through allocation algorithm.

B. PROBLEMS IN TASK SCHEDULING

CLOUD SERVIVE SCHEDULING

The scheduling of cloud service is carried out at user level and system level [11]. If the
problems are raised between the providers and customers due to service provisioning, it is
dealt at user level scheduling while the resource management within datacenter is handled at
system level scheduling.
A datacenter is integrated with many physical machines and the millions of tasks are received
and assigned to the physical machine at datacenter. The performance of the datacenter is
closely related with these assignments. In addition to system utilization other factors such as
the quality of services (QoS), sharing of resources, real time satisfaction (RTS), reliability,
service level agreement (SLA) etc. should be kept into consideration.

USER LEVEL SCHEDULING

To regulate the demand and supply of the cloud resources, the auction-based and market-
based schedulers are suitable schedulers. In the cloud computing environment, the effective
allocation is the market-based resource allocation where the user gets the virtualized
resources as a service.

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

A set of market-oriented task scheduling algorithms to an AuctionNet for heterogeneous

distributed environment is proposed in [12]. Pricing model development using processor-
sharing for the clouds and its applications to composite services with dependency
consideration the development of two sets of profit-driven scheduling algorithms are
proposed in [13]. The provisioning of the services is completely based on the service level
agreement (SLA) in cloud systems. The service level agreement provides a signed contract
between the service provider and the customer containing the terms and conditions of the
agreement including with non-functional requirements such as quality of services, penalties
and obligations in case of breach of trust. Thus, the scheduling strategies along with efficient
resource allocation and multiple SLA parameters are needed. A novel scheduling heuristic
considers multiple SLA parameters for deploying.

STATIC AND DYNAMIC SCHEDULING

In the static scheduling the data are pre-fetched and pipelined to the different stages of the
task execution. The static scheduling has relatively less runtime overhead. In the dynamic
scheduling the job tasks/components are dynamic in nature and are not known before their
execution so the execution time of the task cannot be judged in advance and the tasks’
allocation is carried out on fly as the application executes. An environment of task execution
that avoids scalable static scheduling techniques provides a flexible pricing model to the user
which reduces scheduling overheads for the cloud environment provider, has been discussed
in [11]. In cloud domain which is having three-tier structure, the mechanism of service
request scheduling consisting with service and resource providers along with customers
should fulfill the requirements of consumers and service providers both. An algorithm to the
dynamic feedback is proposed in [13] for the scheduling mechanism of preemptable jobs. In
the cloud environment the fully utilization of resources and the procedure of feedback are
efficiently improved by the preemptable scheduling. This also works well in such conditions
where the contentions of resources are fierce. Because every task on the cloud environment is
quite different from each other so accurate cost of cloud resources cannot be measured by the
traditional way of task scheduling. For the task scheduling, an optimized algorithm and its

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

implementation are introduced in [16] which is based on activity based costing. In the hybrid
cloud environment, an experiment is performed in [17] on different strategies of optimization
for cost-optimal dynamic scheduling.

In the cloud systems, an improved efficient backfill algorithm is proposed in [18] for
achieving the quality of services. This algorithm uses the balanced spiral method (BSM).
This paper has analyzed so many scheduling algorithms which support parallel jobs such as
CBA and EASY. A scheduling algorithm which measures both parameters- computation
performance and cost of the resources is proposed in [19]. This algorithm groups the user
tasks according to the processing capability of a particular cloud resource and pushes the
grouped jobs to the cloud resources and also enhances the computation/communication ratio.
Due to the grouping of the jobs, the computation/communication ratio is optimized by the
communication of resources and coarse-grained jobs.

The tremendous emission of heat, carbon dioxide and energy from the cloud computing
servers results pollution in the environment. Green task scheduling mechanisms which
significantly reduce the pollution by minimizing the above concerned issues are proposed in
[10] and [11]. The scheduling strategies which consider the issues and parameters other than
explained above are discussed in [16] and [17]. Throughout the network, the information
about the execution nodes which are available, is aggregated through the fully decentralized
scheduler and is used to allocate jobs to those execution nodes which are able to execute them
in time.

The study of communication model and realistic network topology proposes Deadline
Reliability Resource-aware (DDR) scheduling algorithm in [13]. An algorithm which is based
on reinforcement learning is proposed in [14] that considers recovery and failure scenarios of
the entities in cloud systems. It makes the scheduling fault tolerable while the utilities are
attained for a long term. A latest framework of job scheduling strategy in tree network has
been proposed in [10].

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

WORKFLOW SCHEDULING
Within a directed acyclic graph the structure of the involved applications is enabled through
its workflow [11], where nodes and edges represent the constituent tasks and their inter-
dependencies to the applications [12]. In the work flow each task may be communicated with
another task and a set of related tasks forms a single workflow. In workflow execution
management, the workflow scheduling is regarded as the key issue. In the cloud systems, a
survey on different workflow scheduling algorithms is presented in [13]. A study of different
scheduling algorithms, issues and problems related to the cloud workflow is discussed in [14]
while [15] covers the study of instance-intensive workflows in cloud systems.

III. RELATED WORK AND SOLUTIONS

A. RESOURCE ALLOCATION ALGORITHM
The optimal use of the cloud resources is driven out very cleverly using the platform-
placement algorithms. Eucalyptus which is available publically has an algorithm that pays a
little attention on the placement of virtual machines within the clout systems. Open Nebula
comes out with an interesting approach for the placement of virtual machines using its rank
assessment by which so many useful policies are implemented. Nimbus presents Workspace
Pilot and Workspace Resource Manager that are liable to take the responsibility for deploying
the jobs and nodes respectively in the cloud environment.

As such, those developers who treat the optimization of resource usage as a very critical
requirement, it can be more attractive to discuss [6]. For IaaS clouds, the scheduling and
resource allocation are key process. These clouds consider the virtual machines as scheduling
units, which are allocated to physical resources of heterogeneous support. To confirm the
resource allocation, Eucalyptus executes Round Robin and First-fit algorithms while Nebula
uses Preemption Scheduling, Advance Reservation and Queuing System algorithms.
Before task scheduling on cloud resources, some characteristics of the cloud systems should
be kept into mind which include:-

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

 Pay per use[1]
 Rapid elasticity
 Location independent resource pooling
 Ubiquitous network access
 On-demand self service

The cloud resources are shared by millions of user via submitting their tasks to the offered
cloud domain. It is a big challenge to schedule these lots of tasks efficiently within the cloud
systems. To do this efficiently, various scheduling strategies are documented in [2] to [10].
These scheduling strategies focus on various factors such as cost matrices which are given by
credit of tasks for a particular resource [2], light weighted VM scheduler based on Backfill
strategy to dispatch jobs, Meta-scheduler based on QoS [3], requirements of QoS [4] and [5],
cloud environment heterogeneity and its workload [8].

In the cloud environment, the allocation of resources and task scheduling decide the number
of systems required for minimizing the total cost. The problems, related to the resource
allocation are continuously occurred due to the highly dynamic nature of cloud computing, as
the availability of servers is addressed while the customers demand for it at the same time.
Thus this analysis concentrates on scheduling algorithms within the cloud systems after
focusing above discussed issues, strategies, characteristics and challenges. For the IaaS
clouds, it’s a key process to schedule resources. Virtual machines are generally used as
scheduling units in these IaaS clouds which have to be allocated to physical resources of
heterogeneous type. In the cloud systems the selection of target resources takes place in
various ways. It can be selected either randomly or by any other means (round robin or
greedy i.e. waiting time and resource processing power based). The scheduling of the job
selection can be done on the basis of SJF, coarse grained task grouping, FCFS or priority
based scheduling etc. Scheduling algorithms are responsible for both, selection of the
executable jobs and the corresponding resources which are to be assigned for the execution of
jobs. As each successive selection strategy comes up with this approach that it will give a
better solution and remove or minimize the prior algorithms’ drawbacks.

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

GREEDY ALLOCATION
Greedy resource allocation algorithms are suitable for such heterogeneous resource
environments which are dynamic in nature and are connected to a scheduler via
homogeneous environment of communication [4]. Greedy approach is one of the best
approaches which are used to solve the problems of job scheduling.

GREEDY RESOURCE ALLOCATION ALGORITHM

B. SOLUTION OF TASK SECHIDULING PROBLEM

Now this paper focuses on developing an optimized algorithm which can help to optimize the
parameters related to the performance of the system. After considering all above mentioned
parameters, this paper will try to optimize the system performance and the users would see
the ultimate improvement in the system performance.

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

IMPLEMENTATION OF ACTIVITY BASED COSTING (ABC) ALGORITHM IN

CLOUD SYSTEMS
Implementation of ABC algorithm can be explained by showing it in a parent-child
relationship structure i.e. tree structure which is represented in figure 3. Using FIFO approach
i.e. according to the tasks’ arrival time, the tasks will be stored in one parent queue. Then the
requested resources and data for their concerned tasks will be checked and sorted into two
disjoint queues- ‘available’ and ‘partially available’. Further in the ‘available queue’, the
dependency and/or independency of the tasks will be checked, according to which again they
will be stored in respective queues. Now, in the queue which is ‘partially available’, there are
some tasks which may need other data centers to provide them data resources. Then these
tasks can be sorted in various queue categories like CT1, CT2, CT 3….and so on. These
queue categories are decided on the basis of tasks’ resources, which they need. For example,
if there are three tasks- task1, task2 and task4. These three tasks need resources from two
locations of data centers- DCL1 and DCL2 then these all three tasks can be stored in one
category. Let this category is CT1. Similarly, the categories CT2, CT3…. and so on can be
made for other similar types of tasks till the partially available queue gets empty.

Now, there are major queues

based on the dependency and
independency of tasks and CT1,
CT2…….CTN. After it, again
three priority-queues are made on
the basis of High, Mid & Low
priorities for each major queue
which are mentioned above. Now
this question arises that how the priority will be decided? The priority is decided by
considering the four major factors- completing time, resources, space needed and profit. The
following formula has been derived to decide the task-priority:-

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

Figure 3: The tree structure of task scheduling

While deciding the priorities of the various tasks, the newly measured priority is checked
with previous ones and then the tasks are placed into the priority-queues. When the placing of
tasks into the queues gets finished, then the tasks from the High priority queues are selected
for their execution. After the execution of the tasks from high priority queues, the tasks of
mid priority queues are shifted to the queues with high priority, and by this way the execution
of the tasks from all the queues gets started.

Again one question arises here that how the selection of the task is sequenced, as there are
many high priority queues? For giving the answer of this question, a very simple logic is used
which is as below:-

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

 Compare the priorities of those all tasks which are at top priority level within their
respective high priority queues
 Then select the task having highest priority and assign space and resources.
 If there are remaining resources, then select next task using same strategy and
assign space and resources.
 Repeat above steps till the end of remaining resources.
 When the allocation of all the resources gets full, then wait till the finishing of any
task and as any task gets finished, choose next task and assign it the resources and
space.

ALGORITHM FOR ACTIVITY BASED COSTING IN CLOUD COMPUTING

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

IV. ANALYSIS OF THE PROPOSED ALGORITHM

A. PERFORMANCE WITH RESPECT TO EXECUTION COST

The tasks select the appropriate resources using the greedy approach and minimize their
execution cost individually. By the proposed algorithm, the execution of tasks results better
than sequential approach as shown in figure 4. As the cloudlets increase, this improvement in
cost increases.
Table 1: Comparison of execution cost

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

The following graph is showing the comparison between the execution cost of sequential and
proposed algorithms. The tasks and execution cost are taken on x-axis and y-axis
respectively:-

Figure 4: Analysis of Execution Cost

B. PERFORMANCE WITH RESPECT TO DEADLINE

The following results show the performance of the cloud system in each trial as the cloudlets
increase:-
Table 2: Comparison of Completion Time

The following graph is showing the comparison between the completion time of sequential
and proposed algorithms. The tasks and completion time are taken on x-axis and y-axis
respectively:-

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

Figure 5: Analysis of completion time

V. CONCLUSION
This paper comes out with new methods of task scheduling and resource allocation
scheduling in cloud domain. It helps in the good understanding of resource allocation and
task scheduling options. With the help of task scheduling, the resources of the cloud systems
are efficiently allocated and managed and optimized.

VI. FUTURE SCOPE

A scheduling policy is always needed for the benefits of the customers and the service
providers both. Each new approach, which may provide the better solution than previous
strategies, is always welcome in research field. As far as the future scope of this work is
concerned, the cost based scheduling policy may be studied more efficiently using this
approach.

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

REFERENCES
[1] QI CAO, ZHI-BO WEI, WEN-MAO GONG International School of Software Wuhan
University Wuhan, China,”. An Optimized Algorithm for Task Scheduling Based On Activity
Based Costing in Cloud Computing”, IEEE 2009.
[2] Jinhua Hu, Jianhua Gu, Guofei Sun, Tianhai Zhao, NPU HPC Center Xi’an, China “A
Scheduling Strategy on Load Balancing of Virtual Machine Resources in Cloud
Computing Environment”, IEEE 2010.
[3] Jeyarani, R., Ram, R. Vasanth, Nagaveni, N., "Design and Implementation of an Efficient
Two-Level Scheduler for Cloud Computing Environment", IEEE, 2010.
[4] GAN Guo-ning, HUANG Ting-Iei, GAO Shuai School of Computer science and
engineering Guilin University of Electronic Technology Guilin, China “Genetic Simulated
Annealing Algorithm for Task Scheduling based on Cloud Computing Environment”, IEEE
2010.
[5] Huang Qi-yi, Huang Ting-lei, "An optimistic job scheduling strategy based on QoS for
Cloud Computing", IEEE, 2010.
[6] M. Malathi, “Cloud Computing Concepts”, IEEE 2011.
[7] Paul, M., Sanyal, G., "Survey and analysis of optimal scheduling strategies in cloud
environment", IEEE, 2012.
[8] Meng Xu, Lizhen Cui, Haiyang Wang, Yanbing Bi, "A Multiple QoS Constrained
Scheduling Strategy of Multiple Workflows for Cloud Computing", IEEE, 2009.
[9] Kuan-Rong Lee, Meng-Hsuan Fu, Yau-Hwang Kuo, "A hierarchical scheduling strategy
for the composition of Algorithms for Preemptable Job Scheduling in Cloud Systems”,
IEEE, 2010.
[10] Qi Cao, Zhi-Bo Wei, Wen-Mao Gong, "An Optimized Algorithm for Task
Scheduling Based on Activity Based Costing in Cloud Computing", IEEE, 2009.
[11] Fei Teng, “Resource allocation and scheduling models for cloud computing”, Paris,
2011.
[12] Han Zhao, Xiaolin Li, “AuctionNet: Market oriented task scheduling in heterogeneous
distributed environments”, IEEE, 2010.

Published By:
 International Journal of Research & Development in
Technology and Management Science –Kailash
Volume - 21| Issue 1 | ISBN - 978-1-63102-445-0 | March 2014
[email protected] | [email protected] | http://journal.rtmonline.in | http://www.ijrdtm.com

Paper Id: IJRDTM – 05272

[13] Boloor, K., Chirkova, R., Salo, T., Viniotis, Y., "Heuristic-Based Request Scheduling
Subject to a Percentile Response Time SLA in a Distributed Cloud". IEEE, 2011.
[14] Luqun Li, "An Optimistic Differentiated Service Job Scheduling System for Cloud
Computing Service Users and Providers", IEEE, 2009.
[15] Qi Zhang, Quanyan Zhu, Boutaba, R., “Dynamic Resource Allocation for Spot
Markets in Cloud Computing Environments”, IEEE, 2012.
[16] Zaman S., Grosu D., “Combinatorial Auction- Based Dynamic VM Provisioning and
Allocation in Clouds”, IEEE, 2012.
[17] Hao Li, Huixi Li, "A Research of Resource Scheduling Strategy for Cloud
Computing Based on Pareto Optimality M×N Production Model",, IEEE, 2011.
[18] Zhongyuan Lee, Ying Wang, Wen Zhou, “A dynamic priority scheduling
algorithm on service request scheduling in cloud computing”, IEEE, 2011.
[19] Laiping Zhao, Yizhi Ren, Yang Xiang, Sakurai, K., “Fault-tolerant scheduling with
dynamic number of replicas in heterogeneous systems”, IEEE, 2011.
[20] Wei Wang, Guosun Zeng, “Trusted Dynamic Scheduling for Large-Scale Parallel
Distributed Systems”, IEEE, 2011.

Published By: