ISSN XXXX XXXX © 2017 IJESC
Research Article
Volume 7 Issue No.3
Clustering Based Task Deployment using Bayes Hypothesis in
Cloud Computing
M.Suvitha 1 , M.Mahil, M.E2
PG Scholar1 , Assistant Professor2
Depart ment of CSE
Govern ment college of Eng ineering, Tirunelveli, Tamilnadu, India
Abstract:
In cloud co mputing, most of the physical host in the data centers are overloaded, that make the cloud data center load imbalanced.
LB-BC (Load Balancing based on Bayes and Clustering) approach is used to find the optimal physical host for task deployment
by achieving a load balancing through a long term proces s. In LBBC approach, for each host constraint value will be calculated
based upon the remain ing resource amount of host and the maximu m requested resource amount. Then the task is deployed to the
physical host, whose constraint value of resource is greater than the constraint value of the task in the data center that are
clustered. It can reduce the failure number of task deploy ments and improve the throughput of data center. It gives the optimal
solution for the task deployment and contains the effective long term load balancing strategy.
Keywords: Task Dep loyment, Load Balancing, Bayes Hypothesis, Clustering, Cloud Co mputing
I. INTRODUCTION
CLOUD co mputing is one of the most promising and valuable
research directions lowing utility computing, grid computing
and distributed computing currently .It provides services of
infrastructure, platform and software for users, and supplies the
on-demand services to users through Internet. Infrastructure as
a Service (IaaS) is the basis of cloud computing. The cloud
data center deploys a large number of physical hosts to supply
user services. Since the remaining resource amount of every
physical host changes constantly, the task cannot be placed
surely to the physical host with the largest remaining resource
amount every time. We assume that the tasks requested by
users are deployed every time to a physical host which is
chosen at random. If the resource amount of the submitted task
is greater than the remaining resource amount of the chosen
physical host, this physical host cannot handle this task. It will
cause a deployment failure of this task. And when the
requested resource amount of a task is similar to the remain ing
resource amount of the physical host in which this task will be
executed, the time used for dealing with this task will
relatively be longer. When a cloud data center receives task
requests constantly, it will make the cloud data center load
imbalanced and thus to result in that computing results cannot
be responded to users timely and effectively. What causes
make the cloud data center’ load so imbalanced that it cannot
provide efficient external services for users? In fact, in order to
ensure the service performance, cloud computing data centers
generally deploy tasks according to the highest load demands
to the corresponding hosts. Therefore, most physical hosts
remain relatively idle in most of the time and it is a waste of
computing resource. At present, the field of load balancing in
cloud data centers is a hot issue. In order to achieve the load
balancing of cloud data centers, it is necessary to choose the
optimal physical host efficiently in the process of deploying
tasks. Most existing work has focused on the problem of how
to achieve the immed iate load balancing within an algorith m
cycle in which the proposed approaches are able to obtain the
optimal task deployment policies for the current. There exists a
limitat ion of overly concentrating on the optimal load
International Journal of Engineering Science and Computing, March 2017
balancing policy fo r the current deployment problem and thus
to make the efficiency decreased and users waiting t ime
prolonged unnecessarily. As is well-known, load balancing is
recognized as a means of providing satisfactory service
performance for users while maximizing the availability and
utilizat ion of the whole cloud system rather than the final
purpose. Further, the satisfactory service performance is easy
to achieve immediately since the availab le amount of
computing resource in cloud date centers is always larger than
the requested amount though this kind of requirements is realtime and strict. Also, as the total resource amount requested by
all users accumulated within a processing cycle has rarely
possibility to approach the current total amount of available
computing resource in a cloud data center, the availability and
performance potential of the cloud data center don’t need to
keep being maximized at all times. Based on the points above,
it is unnecessary to ensure that the optimal load balancing
effect is reached through after each algorith m cycle in real t ime
as long as the whole system has been always tending to a longterm optimal load balancing effect. Based on this motivation,
this paper has proposed a heuristic approach to finding the
optimal physical hosts for tasks deployment by achieving a
load balancing strategy through a long-term algorith m process
and thus to obtain optimal performance. A constraint value is
determined in the light of the resource amount of requested
tasks. That is, each requested task has a constraint value. Then
the physical hosts, whose constraint values of computing
resource are greater than the constraint value of the tasks in the
cloud data center, are clustered. And the physical hosts whose
similarities are within a certain threshold value constitute a set
through clustering. And the physical host set obtained by
clustering methods is the physical host set whose physical
hosts are optimal for deploying the task. Then, the tasks will be
put into physical hosts in the set to deploy. The process of the
clustering of physical hosts in the cloud data center is to find
the optimal physical hosts for deploying tasks. Thus, through
the task deployment strategy, not only the load balancing of
the cloud data center can be achieved, but also efficient
performance of external services can be provided for users.
This paper aims to achieve long-term load balancing of cloud
5228
http://ijesc.org/
data centers and provide efficient performance of external
services. It can be achieved only by deploying requested tasks
into the resource pool of the cloud data center efficiently and
properly, wh ich can imp rove computing efficiency of cloud
data centers, and provide users with cloud data centers good
performance of external service. The proposed load balancing
strategy LB-BC can find optimal physical hosts of task
deployment effectively, and achieve the whole networks load
balancing of cloud data centers chronically. The rest of this
paper are organized as follows: in the second section, the
related work of the cur-rent approaches achieving load
balancing of cloud data centers will be introduced briefly. And
in the third section, first of all, a premise of the proposed
research problem is pointed out, and then the pro -posed
problem is formalized in detail. In the fourth section, the
system architecture is designed first in cloud computing
environment, and then the proposed solution is put forward for
addressing the formalized research problem. In the end of this
section, the design and imp lementation process of the proposed
algorith m are introduced in detail. The fifth section shows
simu lation experiments and results, proving that the proposed
approach has high efficiency. And the conclusion of the whole
paper is made in the sixth section.
II. RELATED WORK
Load balancing has always been a hot research topic of cloud
data centers, and its goal is to ensure that every computing
resource can handle tasks effectively and quickly. Ultimately,
the utilizat ion of resource is improved. Researchers have
proposed a series of static, dynamic and hybrid scheduling
strategies. In addition, there are also some studies using live
migrat ion technology of virtual machine to meet the cloud data
centers’ task requests which include performance requirements
and load limitation. Existing load balancing strategies are
generally divided into two categories: static load balancing and
dynamic load balancing. Static load balancing scheduling
algorith ms are commonly including round robin, weighted
round robin, least connection method, weighted least
connection method and so on. These static algorith ms only use
some static informat ion which cannot reflect dynamic load
changes in the cluster of hosts effectively and they have poor
adaptive ability. Currently, most of open-source IaaS platforms
have utilized static algorithms to conduct the resource
scheduling. For ex-amp le, Eucalyptus platform uses round
robin to assign virtual mach ines to different physical hosts in
sequence to achieve load balancing. In, Wei Qun et al. have
emp loyed the weighted min imu m lin k algorithm, which means
that different weights indicate the performance of the physical
host. Then, the virtual machine will be allocated to the physical
host which has the smallest ratio of the current number and the
weight. The advantage of static scheduling algorithms is that it
is simp le to use. But in the large-scale cloud data centers
whose resource heterogeneity is very strong and user demand
isn’t consistent, the load balancing effect is not ideal. The
proposed LB-BC approach is derived the idea of reducing
unnecessary computation complexity just like the existing
static approaches. However, LB-BC is an efficient approach
which has ability in dynamically achieving the long -term
optimal load balancing effect. Dynamic load balancing is a
NP-co mp lete problem and a classic combinatorial optimizat ion
problem. It is mainly used in the field of distributed parallel
computing, and its main objective is how to distribute the load
more rationally among mu ltiple hosts to avoid some
phenomenon that some computing nodes are overloaded and
some nodes are light-loaded and thus to improve the whole
performance of systems. Additional commun ication over-head
International Journal of Engineering Science and Computing, March 2017
produced in the process of DLB will degrade the system
performance of the dynamic load balancing. And with the
increase of network latency between nodes, the impact of the
additional communicat ion overhead on the performance of
DLB will further increase. Therefore, how to reduce
communicat ion overhead furthest between nodes in the process
of DLB beco mes an important problem which will influence
the performance of DLB. At present, aiming at the problem of
reducing additional commun ication overhead in the process of
DLB, the solution is mainly using greedy algorithms to deal
with it. Load Receiver Strategy (LRS) algorith m which is put
forward in has used the way that the light load is received and
al-located preferentially. Ke Xu et al. have proposed an
efficient double-sided combinatorial auction model to
dynamically achieve optimal goals. In, Sau-M ing Lau et al.
have integrated the two strategies of heavy load priority and
light load priority. They have proposed an adaptive load
distribution algorithm to effect ively reduce communicat ion
overhead of the load balancing process. Utilizing the greedy
algorith ms can solve the problem of load distribution.
However, several algorith ms above cannot meet greedy choice
performance and the nature of optimal sub-structure at the
same time. So these load distribution approaches often obtain
the local optimal solutions. And the effect of solving the
problem of load distribution under certain special
circu mstances is not ideal. Cloud data center cannot reach load
balancing of the entire network. Virtual mach ine migrat ion
placement strategy is a widely used strategy to achieve load
balancing of cloud co mputing data centers currently. VM ware
load balancing solution is DRS (Distributed Resource
Scheduling). When DRS select the physical host for the virtual
mach ine, it will check the load status of each physical host an d
choose the placement solution which can imp rove the overall
load balance degree. And in the process of running a virtual
mach ine, DRS will continuously monitor the load status of the
cluster and use VMware VMotion technology to perform live
migrat ion of virtual machines between different physical
servers. Thus, it can ensure load balancing and efficient
utilizat ion of physical resource of the entire cluster. In Jing Tai
Piao et al. have proposed a network-aware virtual machine
placement and mig ration approach to minimizing the data
transfer time consumption and improving the overall
application performance of the cloud data center. Ho wever,
this approach probably results in a relatively lower utilizat ion
of resource of physical hosts and raises the operation cost of
the cloud data center. Jason Sonneck et al. in have presented a
decentralized affinity-aware migration technique that
incorporates heterogeneity and dynamism in network topology
and job communicat ion patterns to allocate virtual machines on
the available physical resources. The technique monitors
network affinity between pairs of virtual machines and uses a
distributed bartering algorith m, coupled with migration, to
dynamically ad just virtual machine placement such that
communicat ion overhead is min imized and load balancing is
achieved. In Vivek Sh rivastava et al. have presented an
approach to place the virtual machines with strong correlation
of applications intensively. However, the load balancing
problem and cost overheads of cloud data centers weren’t
considered. It only focused on virtual mach ines management to
strengthen the management of cloud data centers and improve
the per-forming efficiency of cloud data centers. Rahman and
Peter Graham have proposed a hybrid approach that combines
static and dynamic provisioning. Its algorith m is to adapt a
good initial static placement of virtual machines in response to
evolving load characteristics using live migration to imp rove
the computing efficiency of cloud data centers. Corentin
5229
http://ijesc.org/
Dupont et al have proposed a flexib le framework for the
(re)allocation of virtual machines in a cloud data center as well
as thus to calculate and work out the best placement of virtual
mach ines, achieving the overall load balancing of cloud data
centers. In addition, Jia Zhao et al have proposed an approach
MOGA-LS, wh ich is a heuristic and self-adaptive multiobjective optimization algorith m based on the improved
genetic algorithm (GA ) and the theory of Pareto optimal
solutions. It aims to achieve dynamic load balancing. In paper,
the proposed LB-BC approach presents a solution to
dynamically achieve the process -based optimal load balancing
with low co mputation complexity and thus obtain the efficient
service performance and computing efficiency on two sides.
The main contributions of this paper are as follo ws:
Introduce the concept of process -based load balancing
in cloud co mputing environ ment, wh ich is in contrast to
the immediate load balancing strategies in the current
literature.
Introduce the model of Bayes theorem on optimal deploy ment
of tasks to work out the probabilit ies of optimal physical hosts.
host is calculated according to the remain ing resource amount
of the physical host and the resource amount users requested.
This approach has optimized the set of physical hosts and
extends the load balancing technique for long term operations.
Li =αLi c+βLi mem ,
Where α+β=1 , and the remain ing attributes values are Li
represent the remain ing computing Power of the physical host
i,Lc
remain ing CPU resource Lmemremaining memory
resource, α weight value of CPU,β weight value of memo ry.
The prior probability of each physical host is calculated
according to the remaining resource amount of the physical
host and the resource amount users requested.
1)
III. PROPOS ED S YSTEM
The concept of process -based load balancing in cloud
computing environment, which fold : to reduce unnecessary
computation complexity and to increase deployment efficiency
while fulfilling the service performance for users. The model
of Bayes theorem on optimal deployment of tasks works out
the probabilities of optimal physical hosts. The cluster-based
task deployment algorith m co mbines with Bayes probabilit ies
to obtain the optimal set of candidate physical hosts.
IV. PROBLEM FORMULATION
IaaS cloud data centers, the system will deploy tasks on the
physical hosts in the resource pool of the cloud data center
when users submit task requests. In general, the cloud data
center will choose the physical hosts at random to deploy
tasks. When the resource amount requested by a task is greater
than the remaining resource amount of the physical host, the
physical host cannot deploy the task. And when the req uested
resource amount is close to the remaining amount of the
physical host, the physical host will have a heavy workload,
causing the decline of its service capacity and computing
power. The decline will result in load imbalance of the cloud
data center and make service efficiency fall. To imp lement the
efficient load balancing strategy to make the cloud co mputing
system efficient is the priority of this project. Load balancing
is the process of distributing the load among various resources
in any system. Thus load need to be distributed over the
resources in cloud-based architecture, so that each resources
does approximately the equal amount of task at any point of
time. The difficulty of the load balancing in data center can be
solved by optimizing task deployment for regular intervals in a
long term. Cloud vendors are based on automatic load
balancing services, which allo w clients to increase the number
of CPUs or memories for their resources to scale with
increased demands. This service is optional and depends on the
clients business needs. So load balancing serves two important
needs, primarily to pro mote availability of Cloud resources and
secondarily to pro mote performance.
LB-BC approach is a heuristic task deployment approach,
which is used to deploy task requests received by the cloud
data center into optimal target physical hosts in the IaaS cloud
computing data center. Its algorithm is what comb ines Bayes
theorem with clustering. The prior probability of each physical
International Journal of Engineering Science and Computing, March 2017
(1)
P (A/B i ) =1-Lmreq/Li
(2)
P (Bi ) =1/m’
(3)
Where A represents the tasks can be executed on some
physical hosts, Bi represent the event that the physical host i is
chosen, Lmreq represent maximu m requested resource amount,Li
is the remaining computing power, m’ represent Physical hosts
in the set. The Posterior Probability value of each physical host
in clustered set as one of its attribute values used in the
clustering process. The posterior probability value of each
physical host is used one of the attribute value in the clustering
process.
(4)
V. S YSTEM ARCHITECT URE
The architecture of the proposed LBBC approach in the cloud
data center environment shows the interaction between LB-BC
and other entities and that it plays a very important role in the
whole architecture. First, the Monitor acquires the informat ion
of resource requested by users’ tasks and the status informat ion
of remaining resource amount (including CPU and memory) of
m availab le physical hosts in cloud data center. By using the
informat ion acquired from Monitor, LB-BC generates the
deployment strategy, which is transmitted to Deployment
Controller whose function is to control and carry out the
deployment of requested tasks. In the end, the task requests
accumulated within a Δt time are deployed to the
corresponding physical hosts in the final optimal set of
physical hosts obtained by the proposed LB-BC approach.
LB_BC  Load B alancing Based on Bayes and Clustering
TD
Task Deployment
Figure.1. Archi tecture Of Lb-Bc Approach
5230
http://ijesc.org/
hosts and 10 v m for each customers. In LB_ BC approach 10%
greater performance than the round robin algorithm. In these
two different datacenters uses 20 hosts and two customers one
is using round robin approach and another one is using LB-BC
approach. Two customers request the same number of v m
requests and the execution times are, LB-BC approach have
lesser time than the round robin approach.
Total execution time(ms)
Total execution time
In Iaas cloud data centers, there are too large numbers of
physical hosts can’t meet situation that the selected physical
hosts can’t meet the resources requirements of requested tasks
and to achieve the best clustering effect through min imizing
the candidate set, we assume that there are m physical hosts in
the cloud data center, and each physical host needs to be
assigned to a constraint value for measuring its remain ing
available co mputing power in the cloud data center according
to (1) and (2). The constraint value Li o f each physical host i in
the cloud data center has been calculated. We define an empty
data center has been calculated. We define an eof the set TR of
task requests is defined as the maximu m requested resource
amount in NPH. The maximu m requested resource amount LM
req of TR can be calculated in terms of formu la.
19.4
19.2
19
18.8
18.6
18.4
18.2
18
17.8
Round robin
LB-BC
Figure.3.comparision of total execution ti me .
Each requested task has a constraint value. Then the physical
hosts whose constraint values of computing resource are
greater than the constraint value of the tasks in the cloud data
center are clustered. Then the physical hosts whose similarit ies
are within a certain threshold value constitute a set through
clustering. And the physical host set obtained by clustering
methods is the physical host set whose physical hosts are
optimal for deploying the task. Then, the tasks will be put into
physical hosts in the set to deploy. The process of the
clustering of physical hosts in the cloud data center is to find
the optimal physical hosts for deploying tasks.
VI. PERFORMANCE ANALYS IS
In this section the proposed LBBC approach compared to the
round robin approach. The follo wing factors are used to
compared the performance evaluation
Number of cloudlets successfully
executed
Number of cloudlets
Figure.2. Fl ow chart of l b-bc approach
The LB-BC is co mpared with the Round Robin approach on
the number of cloudlets to be executed. In these two different
datacenters uses 20 hosts and two customers one is using round
robin approach and another one is using LB-BC approach.
Two customers request the same number of vm requests. The
proposed approach executes maximu m nu mber of cloudlets
than the round robin task deployment approach. In LB-BC
approach spilt the current tasks in maximu m nu mber of
cloudlets than the round robin approach. The cloudlets
deployed the selected physical hosts only so the execution t ime
will be reduced and maximu m nu mber of cloudlets executed in
a short time. In LB_ BC approach 20% greater performance
than the round robin algorith m.
400
350
300
250
200
150
100
50
1.
Co mparison of the total execution time
2.
Co mparison of number of cloudlets successfully
executed.
3.
Co mparison of number of migrations will be done.
LB-BC is compared with the Round Robin approach on the
needed time of processing tasks set. The experimental results
of the two approaches are shown as Figure. The RR approach
has deployed the requested tasks at random on the physical
hosts in the cloud data center. Therefore, the needed time will
also increase. The proposed LB-BC approach will select the
optimal physical hosts set to deploy tasks. The execution time
is smaller than the RR deploy ment approach with the same
number of requested tasks. In this comparison made with 10
International Journal of Engineering Science and Computing, March 2017
0
Round robin
LB-BC
Figure.4.comparison of number of cl oudlets successfully
executed.
The proposed LB_ BC approach compare the number of
migrat ions will be done with the round robin approach. In
these two different datacenters uses 20 hosts and two
customers one is using round robin approach and another one
is using LB-BC approach. Two customers request the same
number of v m requests. In Round Robin approach maximu m
number of migrat ions will be done than the LB_ BC approach.
It increases the execution time of the task deployment than the
proposed algorithm.
5231
http://ijesc.org/
Number of migrations
Number of migrations
14
12
10
[7]. L. Liu, H. Wang, X. Liu, X. Jin, W. He, Q. Wang, and Y.
Chen, ”Green Cloud: A New Arch itecture for Green Data
Center,” Proc. the 6th International Conference Indu stry
Session on Autonomic Co mputing and Communications
Industry Session, pp. 29-38, Jun. 2009.
[8]. Q. Wei, G. Xu, and Y. Li, ”Research on Cluster and Load
Balance based on Linu x Virtual Server,” Proc. Informat ion
Co mputing and Applications, pp. 169-176, 2011.
8
6
4
2
0
Round robin
LB-BC
Figure.5.comparison of number of migrations.
VII. CONCLUS ION AND FUTUR E ENHANCEMENT
LB-BC has utilized Bayes theorem to obtain the posterior
probability values of all candidate physical hosts. LB-BC has
combined probability theorem and the clustering idea to pick
out the optimal hosts set, where these physical hosts have the
most remain ing computing power currently, for deploying and
executing tasks. It makes cloud data centers achieve a long term load balancing of the whole network. Load balancing has
been achieved for host allocation through bayes theorem and
the execution time of handling requests in the data center has
been minimized. Request handling is carried out by node
weight scheduling which reduces the loss of requests in
execution time. A larger scale of simu lation experiments and
prototype will be done. Attempt to implement LB-BC in a real
cloud computing environment and evaluate its performance
and efficiency as well as verify its load balancing effect of
cloud data centers. The migration will be done in a single data
center it will be extend to multip le data centers.
VIII. REFERENCES
[1]. T. Erl, R. Puttini, and Z. Mah mood, Cloud Computing:
Concepts,Technology & Architecture, first ed. Indiana, U.S.:
Prentice Hall,2013.
[2]. S. K. Garg, S. Versteeg, and R. Buyya,”A Framework for
Ranking of Cloud Co mputing Services,” Future Generat ion
Co mputer Systems, vol. 29, no. 4, pp. 1012-1023, Jun. 2012.
[3]. G. Nan, Z. Mao, M. Li, Y. Zhang, S. Gjessing, H. Wang,
and Mohsen Gu izan i, ”Distributed Resource Allocation in
Cloud-based Wireless Multimedia Social Networks,” IEEE
Network Magazine, vol. 28, no. 4, pp. 74-80, Ju l. 2014.
[9]. Y. Di, S. Wang, and X. Sun, ”A dynamic load balancing
model based on negative feedback and exponential smoothing
estima-t ion,” Proc. the 8th International Conference on
Autonomic and Autonomous Systems, pp. 32-37, Mar. 2012.
[10]. W. Chen, Y. Zhang, and Z. Xiong, ”Research and
Realization of the Load Balancing Algorith m for
Heterogeneous Cluster with Dynamic Feedback,” Journal of
Chongqing University, vol. 33, no. 2, pp. 2-14, 2010.
[11]. S. Song, T. Lv, and X. Chen,”A Static Load Balancing
algo-rithm for Future Internet,” Journal of Electrical
Engineering, vol. 12, no. 6, 2014.
[12]. D. Nurmi, R. Wolski, C. Grzegorczy k, G. Obertelli, S.
Soman, L. Youseff, and D. Zagorodnov, ”The Eucalyptus
Open-source Cloud-co mputing System,” Proc. the 9th
IEEE/ACM International Sy mposiu m on Cluster Co mputing
and the Grid, pp. 124-131, 2009.
[13]. M. H. Willebeek-LeMair, and A. P. Reeves ,” Strategies
for Dynamic Load Balancing on Highly Parallel Co mputers,”
IEEE Trans. Parallel and Distributed, vol. 4, no. 9, pp. 979993, Sept. 1993.
[14].T. You, W. Li, Z. Fang, H. Wang, and G. Qu,
”Performance Evaluation of Dynamic Load Balancing
Algorith ms, “TELKOM -NIKA Indonesian Journal of
Electrical Eng ineering, vol. 12, no. 4, 2014.
[15]. J. M. Bah i, S. Contassot-Viv ier, and R. Couturier,
”Dynamic Load Balancing and Efficient Load Estimators for
Asynchronous Iterative Algorithms ,” IEEE Trans. Parallel and
Distributed Systems, vol. 16, no. 4, pp. 289-299, 2005.
[16]. I. Pardines, and F. F. Rivera,”M inimizing the Load
Redistribution Cost in Cluster Architectures,” Proc. the 12th
Euro micro Conference on Parallel, Distributed and Network based Processing, pp. 326-331, 2004.
[4]. R. Yu, Y. Zhang, S. Gjessing, W. Xia, and K. Yang,
”Toward Cloud-based Vehicular Networks with Efficient
Resource Management,” IEEE Network Magazine, vol. 27, no.
5, pp. 48-55, 2013.
[17]. K. Xu, Y. Zhang, X. Shi, H. Wang, Y. Wang, and M.
Shen, ”Online Co mbinatorial Double Auction for Mobile
Cloud Com-puting Markets,” Proc. the 2014 IEEE
International Perfo rmance Co mputing and Co mmunications
Conference (IPCCC), pp. 1-8, 2014.
[5]. G. Xu, Y. Ding, J. Zhao, L. Hu, and X. Fu, ”A Novel
Artificial Bee Colony Approach of Live Virtual Machine
Migration Po licy Using Bayes Theorem,” The Scientific World
Journal, vol. 2013, no. 2013, p. 369209, Sept. 2013.
[18]. S. M. Lau, Q. Lu, and K. S. Leung, ”Adaptive Load
Distribution Algorith ms for Heterogeneous Distributed
Systems with Multiple Task Classes,” Journal of Parallel and
Distributed Co mputing, vol. 66, no. 2, pp. 163-180, 2006.
[6]. J. Zhao, L. Hu, Y. Ding, G. Xu, and M. Hu, ”A Heuristic
Placement Selection of Live Virtual Machine Migrat ion for
Energy-Sav ing in Cloud Co mputing Environ ment,” PloS one,
vol. 9, no. 9, p. e108275, Sept. 2014.
[19]. M. Armbrust, A. Fo x, R. Griffith, A. D. Joseph, R. Kat z,
A. Konwinski, G. Lee, D. Patterson, A. Rabkin, I. Stoica, and
M. Zaharia, ”A View o f Cloud Co mputing,” Co mmunications
of the ACM, vol. 53, no. 4, pp. 50-58, 2010.
International Journal of Engineering Science and Computing, March 2017
5232
http://ijesc.org/
[20]. R. Moreno-Vo zmediano, R. S. Montero, and I. M.
Llorente,”Key Challenges in Cloud Co mputing: Enabling the
Future Internet of Services,” IEEE Internet Co mputing, vol.
17, no. 4, pp. 18-25, 2013.
[21]. A. Gu lati, A. Holler, M. Ji, G. Shanmuganathan, C.
Wald-spurger, and X. Zhu,”VM ware Distributed Resource
Manage-ment: Design, Implementation and Lessons Learned,”
VM ware Technical Journal, vol. 1, no. 1, Mar. 2012.
[22]. J. T. Piao, and J. Yan, ”A Network-A ware Virtual
Machine Placement and Migrat ion Approach in Cloud
Co mputing,” Proc. the 9th IEEE International Conference on
Grid and Coop-erative Co mputing (GCC), pp. 87-92, 2010.
[23]. J. Sonnek, J. Greensky, R. Reutiman, and A. Chandra,
”Starling: Min imizing
Co mmunication
Overhead
in
Virtualized Co mputing Platforms Using Decentralized
Affin ity-aware Migrat ion,” Proc. the 39th IEEE International
Conference on Parallel Pro-cessing (ICPP), pp. 228-237, Sept.
2010.
[24]. V. Shrivastava, P. Zerfos, K. W. Lee, H. Jamjoo m, Y. H.
Liu, and S. Banerjee, ”Application-aware Virtual Machine
Migration in Data Centers,” Proc. IEEE INFOCOM, pp. 66-70,
2011.
[25]. M. Rah man, and P. Graham,”Hybrid Resource
Provisioning for Clouds,” Journal of Physics: Conference
Series, vol. 385, no. 1, p. 012004, 2012.
[26]. C. Dupont, G. Giu lian i, F. Hermen ier, T. Schulze, and A.
Somov, ”An Energy Aware Framework for Virtual Machine
Placement in Cloud Federated Data Centres,” Proc. the 3th
IEEE International Conference on Future Energy Systems:
Where Energy, Co mputing and Co mmunication Meet (eEnergy), pp. 1-10, 2012.
[27]
J. Zhao, Y. Ding, G. Xu, L. Hu, Y. Dong, and X. Fu,
”A Location Selection Policy of Live Virtual Machine
Migration for Power Saving and Load Balancing,” The
Scientific World Journal, vol. 2013, no. 2013, p. 492615, Sept.
2013.
International Journal of Engineering Science and Computing, March 2017
5233
http://ijesc.org/