A Green Model of Cloud Resources Provisioning

Meriem Azaiez

, Walid Chainbi

and Hanen Chihi

National Engineering School of Sousse, University of Sousse, Sousse, Tunisia

Institute of Computer Sciences of Ariana, Ariana, Tunisia

Keywords:

Cloud Computing, Optimization, Scheduling, Green Computing, CloudSim, Genetic Algorithm.

Abstract:

The evolution of network technologies and their reliability on the one hand, and the spread of virtualization

techniques on the other hand, have motivated the use of execution and storage resources allocated by distant

providers. These resources may progress on demand. Cloud computing deals with such aspects. However,

these resources are greedy in energy because they consume huge amounts of electrical energy, which affects

the invoicing of Cloud services which depends on run-time and used resources. The environment is affected

too due to the emission of greenhouse gas. Therefore, we need Green Cloud computing solutions that reduce

the environmental impact. To overcome this Challenge, we study in this paper the relationship between Cloud

infrastructure and energy consumption. Then, we present a genetic algorithm based solution that schedules

Cloud resources and optimizes the energy consumption and CO

emissions of Cloud computing infrastructure

based on geographical features of data centers. Unlike previous work, we propose to optimize the use of Cloud

resources by scheduling dynamically the customers applications and therefore reduce energy consumption

as well as the emission of CO

. The optimal solution of scheduling is found using multi-objective genetic

algorithm. In order to test our model, we extended the CloudSim simulator with a module implementing

the dynamic scheduling of customers applications. The experiments show promising results related to the

adoption of our model.

1 INTRODUCTION

Cloud computing is an emerging ﬁeld which becomes

increasingly popular. But, this technology is identi-

ﬁed as one of the fastest growing consumers of en-

ergy. This consumption of energy will reach in 2020,

more than three times compared to today (Relaxnews,

2010). This problem is caused by the energy con-

sumed by data centers. Indeed, the data centers en-

ergy consumption increases with the number of cen-

ters and with data center workload. This consumption

is ampliﬁed when the cooling infrastructure and aux-

iliary equipment are included which represents more

than 50% of the power consumption (Zhang et al.,

2012).

Another problem with energy is the emission of

greenhouse gas which reached the 2% of the total

amount of CO

emissions in the world (TUAL, 2013)

and will quadruple in 2020 (Thrash, 2012). This prob-

lem brings a huge impact to the environment. There-

fore, a new challenge appears to deal with the energy

consumption and greenhouse gas emissions.

Many studies have addressed the green provision-

ing of resources to reduce energy consumption in

Cloud environment. Most of these works use static

allocation methods such as FCFS (First-Come-First-

Serve) to ensure performance and quality of services

(Calheiros et al., 2011). These strategies are very sim-

ple, but the problem with them is the need of many

available resources. The energetic efﬁciency of these

resource provisioning methods depends on the num-

ber of customers applications. Other methods used

in the Cloud environment are deployed pre allocation

strategies (Nair and Jayarekha, 2012). But the prob-

lem here is the prediction of required resources, which

is difﬁcult.

Unlike theses works, we propose to include green

aspect in Cloud environment to support the Green

Cloud computing. Indeed, the present study al-

lows ﬁnding a solution to the green provisioning by

scheduling customers applications to optimize the use

of Cloud resources and therefore reduce energy con-

sumption as well as the emission of CO

We use the information of the Cloud infrastructure

resources and their relations with energy consumption

and CO

emissions. The main objectives are to min-

135

Azaiez M., Chainbi W. and Chihi H..

A Green Model of Cloud Resources Provisioning.

DOI: 10.5220/0004940701350142

In Proceedings of the 4th International Conference on Cloud Computing and Services Science (CLOSER-2014), pages 135-142

ISBN: 978-989-758-019-2

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

imize these two factors on the one hand, and to re-

duce the cost of services on the other hand. To opti-

mize these objectives, we include our parameters in a

multi-objective genetic algorithm and we execute the

optimal solution in our Cloud which is tested as an

extension of the CloudSim framework.

The remainder of this paper is organized as fol-

lows. Section 2 discusses related work. A detailed

description of our solution is presented in section 3.

Then in section 4, we provide technical details of our

work. Section 5 deals with the experiments and the

results of the simulations which are produced with

comparisons and detailed analysis. Finally, the paper

ends with brief conclusive remarks and discussion on

future studies directions.

2 RELATED WORK

Cloud resources optimization is difﬁcult to meet be-

cause of the uncertainty of future consumer demand

and resource prices. It has been a topic of research in-

terest and development for many years. To address the

growing challenge, techniques from many disciplines

were integrated synergistically. Next, we present the

state of the art of cloud resources optimization meth-

ods.

Cloud computing’s usage-based pricing model

creates an incentive for subscribers to optimize the

utilization of the rented resources. Borovskiy et al.

(Borovskiy et al., 2011) devise a formal approach for

distributing workload among a minimum number of

servers. They model this problem as a linear program-

ming problem and describe two solution approaches.

The ﬁrst one generates a set of candidate blocks and

then composes an optimal partition by solving an in-

teger programming problem. The second approach

solves the set partitioning problem with column gen-

eration technique. The disadvantage of such method

is its difﬁculty to consider the purpose of Cloud re-

sources optimization because of the nonlinear charac-

teristics of users’ demands distribution.

Chaisiri et al. (Chaisiri et al., 2012) propose a

method to optimize Cloud resources cost. The under

provisioning problem can occur when the reserved re-

sources are unable to fully meet users’ demands due

to its uncertainty of the workload distribution. To

address this problem, the authors propose an opti-

mal cloud resource provisioning algorithm based on

stochastic programming model.

Regarding the problem of the description of the

Cloud resources characteristic with nonlinear equa-

tion, some researchers propose the use of a stochas-

tic optimization approach. For example, Li proposes

a model based on stochastic integer programming for

Cloud resources optimization (Li, 2012). He proposes

to address the SLA-aware resource composition prob-

lem. He deﬁnes a stochastic integer programming

model for resource composition and provides an algo-

rithm that implements Grbner based theory to solve

this problem (Buchberger, 2001).To solve the mini-

mization problem of Cloud infrastructure cost, Zhao

et al. developed a deterministic model for resource

reservation planning, using a mixed integer linear al-

gorithm, to generate optimal decisions given ﬁxed pa-

rameters (Zhao et al., 2012). In addition, they pro-

posed a stochastic model of resource rental planning

which explicitly takes into account the uncertainty of

resources and users’ demand in the decision making

process. One major disadvantage of such approaches,

especially in dynamic environments where the opti-

mal solution changes over time, is that the parameter

estimation phase signiﬁcantly delays the implementa-

tion of an optimal solution.

Other researchers use the constraint satisfaction

problem (CSP) approach to solve the problem of

Cloud resources optimization. Van et al. propose

a two-level based architecture that deﬁnes a clear

separation between application speciﬁc functions and

a generic global decision level (Nguyen Van et al.,

2009). They use utility functions to map the current

state of each application for a scalar value that quan-

tify the ”satisfaction” of each application in terms

of its performance targets. These utility functions

are also means of communication with the layer of

global decision which builds a global utility function

including the costs of resource management. The

stage of provisioning of virtual machines has been

separated from the stage of placement of virtual ma-

chines within the global decision layer loop and for-

mulates both problems as constraint satisfaction prob-

lem. Doughertya et al. propose a model driven ap-

proach to optimize the conﬁguration, the energy con-

sumption and the cost of infrastructure for Cloud in-

frastructure self-scaling to create green IT environ-

ments that reduce emissions resulting from the use of

redundant resources unused (Dougherty et al., 2012),

(Dougherty et al., ). They proposed to decompose the

model to four sub-problems to ensure infrastructure

auto-scaling: explaining how virtual machine conﬁg-

urations can be captured; describe how these models

can be transformed in constraint satisfaction problems

for the conﬁguration and optimization of energy con-

sumption; showing how optimal auto-scale conﬁgura-

tions can be derived from these CSPs with a constraint

solver and present a case study of energy consumption

and cost reduction of production of this model-driven

approach. The main drawback of these methods is

CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience

136

there exponential complexity.

Bio-inspired scheduling algorithms are often used

in heterogeneous computing environments. Chaisiri

et al. propose an optimal VM placement algo-

rithm which optimally allocates VM to multiple cloud

providers and follows optimally in advance the re-

sources reservation (Chaisiri et al., 2009). This algo-

rithm minimizes the cost virtual machines hosting in

an environment of multiple cloud providers. Van et al.

show that the ability to automate the provisioning and

dynamic placement of virtual machines, taking into

account both the software-level’ SLA and resource

cost with high-level handles for the monitor to spec-

ify compromise between the two (Van et al., 2009).

The model deﬁnes a support for heterogeneous appli-

cations and workloads including both enterprise on-

line applications with stringent QoS requirements and

batch-oriented CPU intensive applications. It is not

focused on optimization problems that are NP-hard in

their general form.

Kessaci et al. (Kessaci et al., 2011) propose

to optimize the allocation of VM requests using a

Pareto-based meta-heuristic approach. In fact, they

use a multi-objective genetic algorithm and propose

to adopt new mutation and crossover strategies to pro-

duce new generations. The three objectives are con-

sidered in the optimization process: minimize both

energy consumption and CO2 emissions of the cloud

infrastructures and maximize the proﬁt of the suppli-

ers. To formulate the problem, they use a real en-

coding. Each individual is a vector representing the

result of processing a pool of applications received

during the scheduling cycle. The used encoding iden-

tiﬁes three main features: the index of the vector rep-

resents the applications, the value of each cell identi-

ﬁes the VM on which the application will be sched-

uled and the maximum number of application. The

Initialization of the MOGA population is divided into

three steps: read the application pool with the greedy

method (Black, ), initializes one or two elements of

the population by the result of the ﬁrst step and ran-

domly initializes the rest of the population. The prob-

lem of this approach is the use of the greedy algo-

rithm to initialize the population of the MOGA algo-

rithm. Despite their simplicity, greedy algorithms can

be subtle and they are costly and mostly provide a lo-

cal minimum. All of the presented approaches take

into account the optimization of the Cloud resources

but they do not consider the relationship between the

satisfaction of users’ requests and the optimization of

providers’ infrastructures cost. They do not pay at-

tention on how each one of those parts can affect the

other. In fact, the optimization of cost and response

time of clients’ requests are closely related to the

number of available and active resources. Our objec-

tive is to minimize the number of the active resources

that minimize energy consumption. To resolve this

problem, we propose to use a multi-objective opti-

mization. The initialization of the MOGA population

is a real time process. It amounts to read the applica-

tion pool and extracts useful information for the opti-

mization algorithm.

3 THE PROPOSED APPROACH

Cloud computing can be represented in three main

categories which are based on the capacity of abstrac-

tion and the paradigm of services. Thus we have the

SaaS, PaaS and IaaS. In our work, we focus on IaaS

as our goal is Cloud requests scheduling.

Iaas provides processing capacities and storage as

well as network components as standardized services.

These services manage a workload requested by cus-

tomers applications.

3.1 The Mathematical Model

The Cloud model adopted by this study is IaaS with

two-tier architecture as shown in ﬁgure 1. On the one

hand, we have the Cloud services provider and on the

other hand, the customer applications which need ser-

vices.

Figure 1: The architecture of our Cloud model.

The optimization of our objectives, which are the

energy consumption and the CO

emission, is owed

to the exploitation of features offered by geographical

distribution of data centers. Indeed, the parameters of

energy model as well as CO

model are different from

one geographical site to another.

In the Cloud, there are two sources of energy con-

sumption: energy from computing equipment, which

is the energy required for calculation and energy from

auxiliary equipment, which is the energy required for

cooling.

To make the energy equation, we use the model of

CMOS (Complementary Metal-Oxide Semiconduc-

tor) processors by adding two constraints of the prob-

lem which are run time (t

run

) and number of proces-

sors required to run the customer application (nb

Thus, the ﬁnal formula of the energy required for cal-

culation is

cal

= (α f

+ β) ×t

run

× nb

(1)

AGreenModelofCloudResourcesProvisioning

137

The energy performance coefﬁcient (COP) is the

ratio between the produced heat and the energy con-

sumed in the treatment. This factor leads us to deduct

the cooling energy which is

aux

cal

COP

(2)

Therefore, the energy to minimize is the total en-

ergy:

total

= E

cal

+ E

aux

(3)

Regarding the second objective, which is the emis-

sion of greenhouse gas, we use in its formula CO

rate, which depends on geographical locations. The

formula of this second objective is

= E

total

× Rate

(4)

To optimize simultaneously these two objectives,

we use the multi-objective optimization techniques

(Goldberg, 1996).

3.2 The Multi-objective Optimization

A multi-objective optimization problem is to optimize

several objective functions simultaneously. Our opti-

mization problem is deﬁned as follows:

Minimize F(x) = ( f

(x), f

(x)) where both func-

tions to optimize f

and f

are respectively the func-

tion of energy and the function of CO

, x = (x

, , x

)

is the vector of decision variables which are the run

time and the number of required processors, F(x) is

the vector of objectives which will be optimized.

The single optimal solution in the mono-objective

optimization problem is replaced by the concept of

Pareto optimal solutions in multi-objective optimiza-

tion problem. Therefore, the optimal solution is not a

single solution but a set of solutions. To ﬁnd the right

balance of solutions, it is necessary to identify the re-

lation between these objectives. The most used rela-

tion is the relation of Pareto dominance. For this rela-

tion, all efﬁcient solutions are called the Pareto Front.

This set of solutions is a balance where no improve-

ment can be made on an objective without degrada-

tion of at least another objective. So the purpose is to

obtain the Pareto front or converge as much as possi-

ble on this front. Figure 2 shows an example of dom-

inance relation in case there are two objectives to be

maximized.

To solve our problem of multi-objective optimiza-

tion based on Pareto-solving methods, we use the

heuristic algorithms. These algorithms are used to

explore the possible solutions space seeking the best

solution. Among these algorithms, we use multi-

objective genetic algorithm (MOGA).

Different models have been proposed for multi-

objective genetic algorithms including VEGA (Vector

Figure 2: The relation of dominance.

Evaluated Genetic Algorithm), NPGA (Niched Pareto

Genetic Algorithm), NSGA (Non Dominated Sorting

Genetic Algorithm) etc. We adopt in our project, the

NSGA-II because it uses an elitist approach that saves

and injects the best solutions found in previous gen-

erations in the new generations (Melcher, 2007),(Deb

et al., 2002). It uses a sorting procedure based on the

non-dominance which is faster. It requires no param-

eter setting. It uses also a comparison operator based

on a calculation of the Crowding distance. This dis-

tance is calculated from nearby solutions.

In our study, this algorithm makes the scheduling

of the execution of customers applications with re-

source optimization. And to assess the effectiveness

of the algorithm as well as the solution, we calculate

the energy consumption and the emission of CO

after

the execution of the applications.

4 SIMULATION ENVIRONMENT

4.1 CloudSim Extension

Since CloudSim provides an extensible framework

for modeling and simulation of Cloud computing in-

frastructures and services, the present work is in-

cluded in CloudSim as an extension. In fact, to cre-

ate our application, we leverage some basic functions

of CloudSim and we extend some characteristics and

features.

In CloudSim, the management of Cloud resources

and all allocation policies are standard and don‘t con-

sider some characteristics of data centers. So, in the

present work, we use this speciﬁcation to investigate

new model for resource allocation. This new tech-

nique is in accordance with ecological standards.

4.2 Class Diagram

Our application divided into two classes diagram:

The ﬁrst class diagram is presented in ﬁgure 3. In

this diagram we have three types of classes.

To meet the needs of our application, we add two

classes. The ﬁrst one is Localisation. This class con-

tains all locations speciﬁcations of Datacenter. The

CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience

138

Figure 3: Project infrastructure class diagram.

Figure 4: MOGA implementation class diagram.

second class is Process. This class is in charge of all

communications between graphical interfaces, data

access, Cloud conﬁguration and simulation. Also, it

allows to manage all Cloud components.

We modify the class DatacenterBroker to adapt

it to our application. This class denotes a Cloud bro-

ker. It is a mediator between users requests and Cloud

infrastructure. Since DatacenterBroker contains the

process of creation of Cloud infrastructure and man-

agement of resource allocation policies, we change

some details and we implement our new model in this

class.

The other classes are those that we use in this work

to create the environment of our new extension. The

infrastructures of Cloud are represented by the class

Datacenter which manages physical machines. Tech-

nical static properties of datacenters are in Datacen-

terCharacteristics class and desired functionalities of

a storage system are in Storage Class.

The class Host represents a physical machine which

has one or more processing element. Also, it has

memory and bandwidth, managed by allocation poli-

cies, storage capacity and provisioning policy for as-

signment to one or more virtual machines. Each pro-

cessing element of hosts, represented by Pe class, has

a processing capacity. The allocation in virtual ma-

chines depends on physical characteristics of hosts.

User applications, represented by the class Cloudlet,

AGreenModelofCloudResourcesProvisioning

139

are deployed in virtual machines by provisioning pol-

icy CloudletSchedulerTimeShared. Virtual machines

are distributed in Datacenters using the VmAllocation-

PolicySimple class.

The class diagram presented in ﬁgure 4, shows the

classes used to implement MOGA namely NSGAII.

The DatacenterBroker class from the ﬁrst diagram is

used for linking this group of classes to the other and

is used also for running the algorithm.

In this diagram, we have 2 types of classes.

Classes of library for the NSGA-II algorithm that

we use: the core of the algorithm is implemented in

NSGA2 class. In this class, we create an instance of

NSGA2Conﬁguration to specify and store all neces-

sary parameters for genetic algorithm.

The instance of the class NSGA2Event is created in

each generation during the run of the algorithm. In

this class, we store information about the current sta-

tus of the algorithm. We use this class in Assign-

mentNSGA2Listener to extract information and print

it.

Classes that we use to personalize our algorithm:

AssignmentIndividual class implements the class In-

dividual. Its used to describe the populations of ge-

netic algorithm. Each individual present a possible

solution of the resource assignment problem.

For each ﬁtness function, we create a speciﬁc

class which implements FitnessFunction interface.

Since we have 2 ﬁtness functions, we implement

2 classes. The ﬁrst one is EnergyFitnessFunction,

which contains the energy function and the second is

CO2FitnessFunction which contain CO

function.

To observe the performance of our genetic algorithm,

we implement the NSGA2Listener interface using As-

signmentNSGA2Listener class. This class shows ﬁt-

ness function values and other detailed data of the best

individuals found during the run of the algorithm.

When the genetic algorithm ﬁnish, it returns the

best found populations which are non-dominated so-

lutions.

5 RESULTS ANALYSIS

In this section, we present the experiments and the

evaluation that we undertook in order to study the

efﬁciency of our extension of CloudSim in terms of

Cloud computing environments optimization.

We deploy two sets of tests. In the ﬁrst one, we de-

cide on the value of MOGA parameters. Then, in

the next test, we simulate and we analyze the Cloud

environment by taking into account the extension of

CloudSim and we compare the results with the initial

existing approach in CloudSim.

5.1 Parameters of the Algorithm

Genetic algorithms have four parameters. To maxi-

mize the efﬁciency of our algorithm, we have to make

a good choice of its parameters values.

Regarding the ﬁrst parameter which is the size of

the population, it is equal to the number of customers

applications to optimize. By varying the values of

stop criterion of the algorithm, which is the maxi-

mum number of generation, we noticed that the best

individuals are always found before the 1000th gen-

eration. Accordingly, we consider this value as the

maximal generation number and as sufﬁcient to ﬁnd

the solution.

In theory, it was found that the values of the pa-

rameters of evolutionary algorithms vary in a speciﬁc

interval. Indeed, former studies (Mais et al., ) have

shown that the best results are achieved by a value

of crossover probability between 0.45 and 0.95, and a

value of mutation probability between 0.01 and 0.005.

To ﬁx these two parameters, we kept the same struc-

ture of the Clouds environment, the same resources

and the same customers applications. Then, we vary

the two remaining parameters of MOGA to choose

the values that give the best results.

To test the effect of the different values of mu-

tation probability on our Cloud environment, we as-

sume that the value of crossover probability is 0.9

which is a predetermined value. Also for the test of

crossover probability, we use the predetermined value

of mutation probability which is 0.05.

The evaluation of tests show that the variation

of probability values of the two parameters in the

theoretical range gives almost stable results. Hence,

we keep the predeﬁned value of crossover probability

and we choose the upper border of the interval of

mutation probabilities. Consequently, values of

0.9 as crossovers probability and 0.01 as mutations

probability may give a satisfactory result. We keep

these values to simulate the different test cases.

5.2 Simulation Results

The purpose of this experiment is to discuss the per-

formance analysis of our approach compared with

static allocation method in terms of the efﬁciency of

resource utilization for the same workload. The static

allocation method used is the method deployed in the

framework CloudSim before extension.

In these experiments, we calculate 2 metrics. The

ﬁrst one is the total energy consumption by the phys-

ical resources of data centers caused by customer ap-

plications workloads which is presented in ﬁgure 5.

CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience

140

The second one is the total emission of greenhouse

gas caused by energy consumption of data centers

which is shown in ﬁgure 6.

Figure 5: Energy consumption.

Figure 6: The emission of CO

To demonstrate the amount of this optimization,

we calculate the percentage of decrease of energy

consumption as well as the emission of CO

. The re-

sult of this demonstration is presented in ﬁgure 7.

The experimental results show that our proposed

approach provides better results compared to results

obtained by static method of resource allocation. In-

deed, from these results we can conclude that for the

use of our solution on Cloud environment, the energy

consumption is reduced to 50% and CO

emissions

up to 60%. This reduction depends strongly on the

characteristics of available resources and the amount

of applications which will be run on the Cloud. The

obtained results are due to the efﬁcient scheduling of

customers applications, and to the reduction of the use

of energy-intensive resources.

Figure 7: Percentage of decrease.

6 CONCLUSION

To address the problem of energy efﬁciency, we have

proposed in this paper a new approach for a Cloud

computing environment that schedule resources allo-

cation based on energy optimization functions. More

precisely, the presented work has optimized the re-

sources in the Cloud, and has minimized the rate of

energy consumption and CO

emissions by the man-

agement of Cloud computing resources and the effec-

tive choice of our genetic algorithm parameters. The

experimental results have shown that the proposed

approach leads to a signiﬁcant reduction in energy

consumption and CO

emissions in comparison with

static techniques of resource allocation.

This study opens more challenges to reduce the

impact of new technologies on the environment, en-

courage the ecosystems, and support energy efﬁ-

ciency.

In future work, we plan to integrate in our project,

an agent system in order to make the Cloud environ-

ment auto-adaptive. A study is underway in order to

fulﬁll this objective.

REFERENCES

Black, P. Greedy algorithm. dictionary of algorithms and

data structures (online), us national institute of stan-

dards and technology, february 2005.

Borovskiy, V., Wust, J., Schwarz, C., Koch, W., and Zeier,

A. (2011). A linear programming approach for opti-

mizing workload distribution in a cloud. In CLOUD

COMPUTING 2011, The Second International Con-

ference on Cloud Computing, GRIDs, and Virtualiza-

tion, pages 127–132.

Buchberger, B. (2001). Gr

obner bases: A short introduc-

tion for systems theorists. In Computer Aided Systems

TheoryEUROCAST 2001, pages 1–19. Springer.

AGreenModelofCloudResourcesProvisioning

141

Calheiros, R. N., Ranjan, R., Beloglazov, A., De Rose,

C. A., and Buyya, R. (2011). Cloudsim: a toolkit for

modeling and simulation of cloud computing environ-

ments and evaluation of resource provisioning algo-

rithms. Software: Practice and Experience, 41(1):23–

50.

Chaisiri, S., Lee, B.-S., and Niyato, D. (2009). Opti-

mal virtual machine placement across multiple cloud

providers. In Services Computing Conference, 2009.

APSCC 2009. IEEE Asia-Paciﬁc, pages 103–110.

IEEE.

Chaisiri, S., Lee, B.-S., and Niyato, D. (2012). Opti-

mization of resource provisioning cost in cloud com-

puting. Services Computing, IEEE Transactions on,

5(2):164–177.

Deb, K., Pratap, A., Agarwal, S., and Meyarivan, T. (2002).

A fast and elitist multiobjective genetic algorithm:

Nsga-ii. Evolutionary Computation, IEEE Transac-

tions on, 6(2):182–197.

Dougherty, B., White, J., and Schmidt, D. C. Model-driven

conﬁguration of cloud computing auto-scaling infras-

tructure.

Dougherty, B., White, J., and Schmidt, D. C. (2012).

Model-driven auto-scaling of green cloud computing

infrastructure. Future Generation Computer Systems,

28(2):371–378.

Goldberg, D. (19 janvier 1996). Algorithmes g

etiques:

exploration, optimisation et apprentissage automa-

tique. Addison-Wesley France, SA, Kluwer Academic

Publisher.

Kessaci, Y., Melab, N., Talbi, E.-G., et al. (2011). Optimi-

sation multi-crit

ere pour l’allocation de ressources sur

clouds distribu

es avec prise en compte de l’

energie. In

Rencontres Scientiﬁques France Grilles 2011.

Li, Q. (2012). Applying stochastic integer programming to

optimization of resource scheduling in cloud comput-

ing. Journal of Networks, 7(7):1078–1084.

Mais, H.-R., Bloch, C., RAMDANE-CHERIF, W., and

Chatonnay, P. Diff

erentes op

erateurs

evolutionnaires

de permutation: s

elections, croisements et mutations.

Melcher, J. (11 September 2007). Jnsga2 1.1 - a free java

library for the nsga-ii algorithm.

Nair, T. and Jayarekha, P. (2012). Pre-allocation strate-

gies of computational resources in cloud computing

using adaptive resonance theory-2. arXiv preprint

arXiv:1206.0419.

Nguyen Van, H., Dang Tran, F., and Menaud, J.-M. (2009).

Autonomic virtual resource management for service

hosting platforms. In Proceedings of the 2009 ICSE

Workshop on Software Engineering Challenges of

Cloud Computing, pages 1–8. IEEE Computer Soci-

ety.

Relaxnews (02-04-2010). Data centers et cloud computing

mettent trop de co2 selon greenpeace.

Thrash, B. (27-02-2012). The green data center opportunity.

TUAL, M. (14-04-2013). Data centers: la donne colo.

Van, H. N., Tran, F. D., and Menaud, J.-M. (2009). Sla-

aware virtual resource management for cloud infras-

tructures. In Computer and Information Technology,

2009. CIT’09. Ninth IEEE International Conference

on, volume 1, pages 357–362. IEEE.

Zhang, Y., Wang, Y., and Wang, X. (2012). Electricity

bill capping for cloud-scale data centers that impact

the power markets. In Parallel Processing (ICPP),

2012 41st International Conference on, pages 440–

449. IEEE.

Zhao, H., Pan, M., Liu, X., Li, X., and Fang, Y. (2012). Op-

timal resource rental planning for elastic applications

in cloud market. In Parallel & Distributed Processing

Symposium (IPDPS), 2012 IEEE 26th International,

pages 808–819. IEEE.

CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience

142