A Novel Green Service Level Agreement for Cloud Computing using

Fuzzy Logic

Awatif Ragmani, Amina El Omri, Noreddine Abhgour, Khalid Moussaid and Mohamed Rida

Department of Mathematics and Computer Science, Faculty of Sciences Ain-Chock, Hassan II University of Casablanca,

B. P. 5366 Maarif, Casablanca, Morocco

Keywords: Cloud Computing, Service Level Agreement, Energy Efficiency, Virtual Machine Consolidation, Fuzzy

Logic.

Abstract: Cloud computing has several features including elasticity and economy of scale that have allowed it to find

several uses in scientific, economic and industrial fields. However, the Cloud model relies heavily on the

architecture of the data centers that host the Cloud services. To date, these data centers consume huge

quantities of electrical energy in order to ensure the operation of both computer equipment and auxiliary

equipment such as cooling systems. In order to reduce the environmental and economic impact of this energy

consumption, several initiatives have emerged especially in the context of Green computing. Through this

article, we propose a Cloud architecture that includes a service level agreement negotiation module based on

the concept of fuzzy logic. The proposed solution aims to introduce a virtual machine consolidation policy in

order to complete a global three-tier architecture. The final solution includes different modules of load

balancing and scheduling based on fuzzy logic, metaheuristic and Map reduce algorithms in order to optimize

both energy efficiency, response time and cost of Cloud services.

1 INTRODUCTION

Cloud computing is an innovative concept that has

quickly gained an important place in the scientific and

industrial fields. The Cloud services model is based

on the use of a legal and business agreement which

manage the relationship between the customer and the

service provider (Jamshidi et al., 2014). his contract

is commonly referred to as service level agreement

(SLA). This agreement sets, among other things, the

pricing terms and the level of quality of service

(SLA). This agreement sets, among other things, the

pricing terms and the level of quality of service

(Wieder et al., 2011). The main principle of the Cloud

is based on the use of virtual resources hosted on the

different physical servers that are grouped within data

centers. The administration of resources within the

data centers relies on different tasks including the

creation of virtual machines (VM), the allocation of

virtual machines within physical servers, the

migration of VMs and the removal of VMs at the end

of users’ requirements. The process of migrating a

virtual machine from one node to another could be

motivated by energy efficiency reason through the

consolidation of multiple virtual machines within a

single server and turning off the other servers that are

no longer used. The virtual machines consolidation

cloud have other motivations such as enhancing the

performance or reliability of the system (Furht and

Escalante, 2010; Josyula, 2012).

Despite the multiple advantages of Cloud

services, the expansion of data centers which host

those services has induced a negative impact on

energy consumption. In order to limit this energetic

effect on the environment and to reduce the

operational cost, several research studies have tackled

the issue of energy efficiency at different levels.

Those research studies could be classified according

to three themes that include the improvement of the

hardware used, the research papers that propose

scheduling and load balancing strategies and the

works that propose hybrid solutions combining the

hardware and software aspect (Banerjee et al., 2013;

Horri et al., 2014; Marotta et al., 2018). One of the

commonly used methods to improve the energy

efficiency within data centers is the dynamic voltage

and frequency scaling (DVFS) method. This concept

is based on the use of the cubic relationship between

CPU operating frequency and power consumption.

This technique is applied both at the level of computer

658

Ragmani, A., El Omri, A., Abghour, N., Moussaid, K. and Rida, M.

A Novel Green Service Level Agreement for Cloud Computing using Fuzzy Logic.

DOI: 10.5220/0006815506580665

In Proceedings of the 8th International Conference on Cloud Computing and Services Science (CLOSER 2018), pages 658-665

ISBN: 978-989-758-295-0

servers as well as network switches. In order to

improve the DVFS technique, a second method called

dynamic power management (DPM) has been

proposed. This technique allows switching off or

putting into standby mode one or more components.

Particularly, the authors of (Kliazovich et al., 2013)

highlight the strong positive impact on energy

efficiency of shutting down an inactive computer and

network hardware servers. However, they issue a

reservation as to the inability to wake up a sleeping

server in a reasonable time which could degrade the

quality of service. In other words, the establishment

of a well-organized strategy for energy efficiency

does not rely merely on efficient physical machines.

The overall policy must take into account all the

mechanisms used within the Cloud environment.

Several research papers have in turn addressed one

aspect of energy efficiency within the Cloud

environment. The best approaches remain the one that

combines the use of energy-aware servers and smart

virtual machines placement. The present paper aims

to introduce a green service level agreement model

taking into account the energy efficiency aspect. The

main contribution of this article is the definition of a

Cloud architecture able to track SLA violations in

terms of performance and energy consumption in

order to guarantee the energy efficiency of the Cloud

data center while respecting the quality of service

levels. The rest of the paper include section 2 which

summarizes the main research papers that have

oriented the present article. Section 3 introduces the

key concept behind the suggested solution. Section 4

highlights the proposed solution and section 5

concludes the present paper.

2 RELATED WORK

Optimizing energy consumption in the Cloud cannot

be accomplished without the active and responsible

participation of customers and providers.

Particularly, several research studies have tackled the

implementation of a green service level agreement

able to guarantee a better respect for sustainable

development and environmental standards within

data centers. The following section introduces some

related work of SLA implementation, including green

SLA. The article (Goyal et al., 2016) tackles the

problem of minimizing energy consumption by

relying on the definition of a service level agreement

(SLA) oriented green computing. The authors

proposed a resource allocation algorithm via a Green

SLA agreement model. The article includes Cloud

service negotiation policies that provide the best

energy efficiency for users. In addition, the authors of

(Moreno and Xu, 2011) defined a resource allocation

model in a real-time Cloud environment. Thanks to a

better estimation of the requested resources, the

system minimizes energy waste. In addition, the

system limits the negative effects of energetic

optimization policies on the SLA requirements

through dynamic remuneration. The tests carried out

by the authors underline the major role of the policy

of virtual machine migration on the performance of

the system.

The authors of the research paper (Djemame et al.,

2011) emphasized the role of the services level

agreement in expanding the commercial uses of the

Grid concept. In order to improve the efficiency of

SLAs by minimizing the probability of failure, the

authors introduced an SLA trading mechanism taking

into account the risk assessment aspect. The proposed

model makes it possible to rate an SLA based on the

probability of failure which has increased the

reliability of the system.

In the article (Verma, 2004), the authors present a

general view of the specificities of service level

agreements in IP networks. Their approach includes a

study of the specific components of a service level

agreement by specifying all three and identify three

axes common to all SLAs in IP networks. Finally, the

authors introduce the possibilities of dynamic SLA

implementation that are more adapted to the IP

network context than static SLAs. The authors of the

research paper (Carlsson and Fullér, 2013) have

defined a probabilistic and possibilistic risk model

that is used to measure the risk of an SLA in a grid or

Cloud computing environment. The authors relied on

a predictive probabilistic approach and the definition

of an increase in the number of failures for the

management of computing resources. Moreover, the

proposed model makes it possible to estimate the

possibility of future failure of a node based on the

concept of fuzzy non-parametric regression.

Another study (De Marco et al., 2015)

emphasized the role of the service level agreement as

a guarantor of the commitments negotiated by both

parties, including the customer and the supplier,

during the period of validity of the service. The

authors proposed addressing the problem of breaches

of the SLA by cybercriminals without both parties

realizing it. The solution proposed by this paper is to

guarantee a better control of the respect of the SLAs

by analyzing the log files. The final model relies on a

mechanism for automatic detection of SLAs

violations. Regarding the effectiveness of fuzzy logic,

the authors of the research papers (Jamshidi et al.,

2015, 2016) have proposed an automatic scaling

A Novel Green Service Level Agreement for Cloud Computing using Fuzzy Logic

659

approach that allows cloud applications to maintain

an adequate level of resources in order minimize costs

and monitor performance during phases of high

demand. The authors introduced a hybrid cloud

controller using both fuzzy logic concepts and Q-

Learning. This combination has proved its efficiency

thanks to the quality of the update of the control rules.

In summary, several studies have been interested

in the evaluation of energy efficiency in data centers

and have proposed different techniques that are

applicable at the level of hardware, scheduling, and

load balancing. On the other hand, different research

papers have examined the strength that an SLA could

play in optimizing the performance of Cloud services

and improving the security and reliability of services.

Through the present paper, we will assess the

possibility of using a SLA negotiation module for

reducing energy consumption.

3 PRELIMINARIES

This section introduces the main concepts taken into

account when developing the Cloud computing

architecture presented in this article, including the

notion of three-tier architecture and service level

agreement.

3.1 Service Level Agreement

The establishment of a service level agreement is a

complex process due to the number of stakeholders

and the multitude of constraints to be included. The

quality of service levels offered in a SLA must not

exceed the actual need of users or be below their

expectations. Service level agreements are geared

more towards the accuracy of the expected service

levels rather than specifying the mode of service

operation. The common description of a SLA is a

support which includes the details of the defined

services, the levels of quality of service, the

measurement indicators and the commitments of the

supplier and the user on the clauses to be respected.

Any overtaking by one party or the other is

considered as a violation of the SLA. The service

level agreement acts as an arbiter in the relationship

between the Cloud services provider and other parties

who may be a service user or negotiator.

In short, the SLA provides an objective and

measurable translation of service level objectives

such as performance, availability and pricing plan. In

addition, the SLA describes the remedies and

penalties in the event of litigation or contravention.

The service level agreement has been widely used in

the field of IT applications and services including

web services in order to describe the evaluation

criteria and key performance indicators. The SLA

allows each stakeholder to monitor compliance with

commitments which contributes to the establishment

of an effective partnership between user and service

provider (Aljoumah et al., 2015). In particular, SLAs

have become very important in terms of Cloud

computing because of the obligation to measure the

quality of Cloud services and to draw up periodic

performance reports. However, the complexity and

scope of Cloud models make it difficult to identify

causes of service disruption (Verma, 2004). Finally,

as described in Figure 1, a typical SLA follows six

major phases to be accomplished that include:

 the creation of a service model and its

attachment to a suitable SLA;

 negotiating terms of the SLA defined;

 provisioning and implementation of the

requested service;

 the execution of the service;

 the monitoring and application of the necessary

corrections during the execution phase;

 the closing and decommissioning of the

achieved or terminated service.

3.2 Data Center Topology

In recent years, the three-tier architecture (see Figure

2) including servers and switches has become the

most applied layout within a data center. This

architecture consists of three levels, the main level

which corresponds to the root of the tree, the level of

aggregation that takes care of the routing, and the

level of access that regroups all the servers in the data

center. This new architecture is an evolution of the

previous version which included only two levels,

without the level of aggregation. But this two-tier

configuration was soon overtaken by the bandwidth

constraint that limited the capacity of hosts to 5,000

units depending on the category of switches used.

Considering the size of the server pool in today's data

center, which are of the order of 100,000 hosts, it

becomes essential to switch to the three-level

architecture. Note that the three-tier architecture

organizes the servers in racks that are interconnected

using a 1 Gbit Ethernet link (Kliazovich et al., 2013).

CLOSER 2018 - 8th International Conference on Cloud Computing and Services Science

660

Figure 1: A service level agreement life cycle schema.

Figure 2: A typical three-tier data center topology.

3.3 Fuzzy Logic Controller

The proposed architecture is based on the use of the

different module in charge of the various process

including the load balancing, scheduling, and the

SLA violation. In this article, we tackled the SLA

violation detection module which uses the concept of

fuzzy logic. The fuzzy logic is defined as being an

extension of Boolean logic. This concept was

introduced by Lotfi Zadeh in 1965 based on the

theory of fuzzy sets, which is a generalization of

classical set theory. This theory uses the notion of the

degree of truth of a condition which allows a

condition to be situated in another state other than

false or true (see Figure 3). This reasoning is based on

the fact that it is closer to human reasoning and can

cover a broader set of options without mentioning the

fact that uncertainties are taken into account(Chen

and Pham, 2001). The definition of the fuzzy is

represented by a function which describes the degree

of the membership. This mathematical function can

get different values between 0 and 1. In brief, a fuzzy

set A is defined as pairs , 









 and





, 



| ∈ , 









∈



0,1



. One of the

essential components in fuzzy logic is the

defuzzification stage. Several methods have been

propped, but the most applied remain Mamdani and

Larsen (Chen and Pham, 2001). As depicted in Figure

4, the proposed Cloud model is based on a three-tier

architecture. The authors of (Jamshidi et al., 2014,

2015) have applied the fuzzy logic in order to define

a very interesting self-learning Cloud controllers.

They introduced that one main aspect of fuzzy logic

is the fuzzy inference which is the procedure of

mapping a list of control inputs to a list of control

outputs by applying fuzzy logic rules. This mapping

is used to describe the controlled output. The

significant advantage of fuzzy controllers still the

definition of the control system for problems which

cannot be characterized by precise mathematical

formulas because of the non-linearity of the studied

system. Hence the interest of fuzzy logic which offers

an alternative solution to this problem by defining an

approximation based on the use of the knowledge in

a comparable manner to the human reasoning.

4 PROPOSED ARCHITECTURE

As noted in previous sections, the environmental

aspect is becoming increasingly important in the

design and operation of data centers. In this section,

we introduce a Cloud architecture that is based on the

concept of fuzzy logic which has been applied to the

service level agreement module in order to detect the

energy consumption level violations. The proposed

module aims to implement a green SLA in order to

ensure energy-aware in a data center by consolidating

virtual machines on a minimal number of servers.

Figure 4 schematizes the proposed architecture for

Cloud environment. The proposed architecture is

composed of several multi-objective modules that

aim at the overall purpose to guarantee the best

response time, the optimal energy efficiency and the

lowest service cost. Particularly, this article focuses

on the description of the module relating to the

detection of SLA violations concerning the level of

energy consumed. This module has been defined

using fuzzy logic for two main reasons. Firstly, the

fuzzy logic allows us to have an intelligent

monitoring of the level of energy consumption and

secondly this additional control does not increase

considerably the response time of the Cloud services

because of the simplicity of the fuzzy controller. The

first level includes the main controller, whose mission

is to centralize the management of the data center.

After receiving requests from users accompanied by

their co-signed SLA specifying the commitments of

A Novel Green Service Level Agreement for Cloud Computing using Fuzzy Logic

661

each party, the main controller generates virtual

machines in charge of processing the users’ requests

according to a process that includes the steps of

scheduling, mapping, load balancing and in case of

necessity virtual machine migration. This process has

already been detailed in our previous work (Ragmani

et al., 2016). The main controller is supported by the

secondary controller, whose function is, on the one

hand, to reduce the workload of the main controller

by monitoring the state of the network and

maintaining a database describing the status of the

hosts and their workloads at intervals T. The

secondary controller informs the primary controller

only in the event of a change of state. Lastly, the

secondary controller provides the backup and the

relay of the main controller in case of failure. The

second level includes the regional load balancer.

Following earlier studies, we have shown that the

response time is greatly improved when assigning a

user to a node close to him geographically. Each load

balancer is connected to a set of nodes which

constitutes the third level of the proposed

architecture. Each node has a technical capability

and can create a set of virtual machines which could

respond to users’ requests. In addition to the

components of the three-tier architecture, the

proposed model includes a module for detecting SLA

violations in terms of performance and energy

consumption. This module is based on the concept of

fuzzy logic. The choice of control technique focused

on fuzzy logic for two main reasons. First, the fuzzy

logic is fast and simple to implement which

contributes to maintaining the performance of the

data center and provides an additional level of

control. Secondly, fuzzy logic makes it possible to

take into account uncertainties and complex cases

where an arbitrary choice must be made between

response time (RT) and power consumption (PC)

based on a SLA violations detection indicator

(SLAVDI).

Figure 3: Structure of the fuzzy logic controller.

In other words, the fuzzy logic module control rules

rely on logical rules in the form of If-Then clauses.

These rules are defined by numerical measurements

or knowledge base. In short, the control performed by

the module of the fuzzy logic follows four steps:

 the description of the energy efficiency and

performance control rules based on the co-

signed SLA content;

 the definition of membership functions on the

knowledge base recorded by the system;

 the application of the logical rules defined by

the system administrator via the main

controller;

 the application of the defuzzification module.

Thus, at the time of the co-signature of the SLA, a set

of rules is generated based on the commitments of

both parties. These rules, described below as fuzzy

controls, will be used to assess compliance with SLA

clauses:

 If (power consumption is low) and (response

time is very fast) then (SLAVDI is very good).

 If (power consumption is medium) and

(response time is very fast) then (SLAVDI is

good).

 If (power consumption is high) and (response

time is fast) then (SLAVDI is medium).

 If (power consumption is very high) and

(response time is medium) then (SLAVDI is

bad).

 If (power consumption is very high) and

(response time is very fast) then (SLAVDI is

bad).

 If (power consumption is very high) and

(response time is slow) then (SLAVDI is very

bad).

 If (power consumption is low) and (response

time is slow) then (SLAVDI is bad).

Based on the value of the SLAVDI, the main

controller could activate the virtual machine

migration process in order to decrease the overall

power consumption of the data center. As depicted in

Figure 5 and 6, the total power consumption of a data

center increases considerably if we apply a virtual

machines migration policy. These figures correspond

to the energy consumption recorded during two

simulations carried out within the CloudReport

(Teixeira Sá et al., 2014) and CloudSim platform

using one data center and 5 consumers.

Inputs

• Response

time

•Power

consumption

Fuzzifier

Inference

engine

If-Then

statements

Defuzzifier

Output

SLAVDI

CLOSER 2018 - 8th International Conference on Cloud Computing and Services Science

662

Figure 4: The proposed SLA architecture for Cloud computing model.

Figure 5: Power consumption with VM migration enabled.

Figure 6: Power consumption without VM migration.

5 SIMULATIONS AND RESULTS

In order to achieve the various simulations (see Table

1), we use the CloudReport simulator (Teixeira Sá et

al., 2014). This toolkit is a graphic simulator based on

the concept of Cloud. CloudReport offers the

opportunity to disconnect the programming and

experimentation roles. This simulator allows the

accomplishment of repeatable trials by changing

simulation parameters and modelling Cloud

architecture by defining one or more data center

which includes multiple hosts. Each server may host

numerous virtual machines. CloudReport parameters

could be adapted in order to perform different

simulation in economic and easy manner. Those

parameters include the number of VM, number of

hosts, customers, VM scheduling, broker policy, data

size, power model and CPU utilization model.

Particularly, the service broker policies parameter

allows the control of the data center which achieves

the user’s request. The CloudReport use by default a

load balancing policy based on round-robin

algorithm. The accomplishment of the predefined

scenarios (see Table 1) has been planned in two

stages. For the first time, we achieve different

experimentation on CloudReport. As depicted in

Figures 7 and 8, the power consumption is highly

influenced by CPU utilization and scheduling model.

Indeed, in the case of applying the stochastic

configuration for CPU, RAM, and bandwidth

utilization model, the achieved result is better than the

case of applying the full configuration. All these

conclusions are used to update the knowledge base

used by the fuzzy controller in order to reduce the

value of SLAVDI (see Figure 9). In a second time, we

apply the fuzzy logic toolbox of Matlab to calculate

and predict the SLAVDI indicator per scenario (see

Figure 10). In summary, we apply 3 scenarios by

modifying the CloudReport parameters. Each

scenario has produced a report and raw data which

has been used to finalize the fuzzy module for service

level agreement detection.

A Novel Green Service Level Agreement for Cloud Computing using Fuzzy Logic

663

Table 1: CloudReport simulations.

Customer Cloudlet Data center Provider Customer

Scenarios

Number VM

RAM/VM

Bandwidth

CPU utilization

model

RAM

utilization

model

Bandwidth

utilization

model

File size

Number host

VM migration

Power model

Maximum

power

Resources

utilization (Max

CPU (MIPS)

Power

consumption

Average finish

Execution time

(ms)

Resources

utilization (Max

CPU (MIPS)

S1 10 512 10 Full Full Full 500 4 Enabled Linear 250 25 25 1700 90

S2 10 512 10 Stochastic Stochastic Stochastic 500 4 Enabled Linear 250 22 17 1700 60

S3 14 512 10 Stochastic Stochastic Stochastic

500 3

Enabled Linear

250 28 32 1700 60

Figure 7: Provider resources utilization during scenario 1.

Figure 8: Provider resources utilization during scenario 2.

Figure 9: The fuzzy logic rules.

Figure 10: Response surface view.

6 CONCLUSIONS

This article has proposed a three-level architecture

that includes a control module based on the concept

of fuzzy logic that can detect SLA violations in terms

of energy consumption. The main role of the

introduced module and the deceleration of the virtual

machine migration procedure in order to reduce the

energy consumption while maintaining a level of

quality of service. Thus, the role of fuzzy logic is

based on the calculation of an SLAVDI performance

indicator that takes into account both the level of

energy consumption and the response time of the

system. CloudSim and CloudReport simulations

demonstrate the positive impact of an efficient and

fast migration policy on the energy consumption of

the data center. In a second step, we plan to finalize

the implementation of the proposed solution within a

Cloud model and to upgrade the fuzzy controller in

order to be autonomous by using machine learning

algorithms. The choice of fuzzy logic was justified by

the efficiency and rapidity factors. Finally, we aim to

finalize a comparative study of the proposed method

regarding other concepts such as (Arabnejad et al.,

2016; Buyya et al., 2011; Jamshidi et al., 2015).

CLOSER 2018 - 8th International Conference on Cloud Computing and Services Science

664

ACKNOWLEDGEMENTS

The authors would like to present their gratefulness to

the anonymous referees for their valuable suggestions

which have greatly contributed to improving the

content of this paper.

REFERENCES

Aljoumah, E., Al-Mousawi, F., Ahmad, I., Al-Shammri,

M., Al-Jady, Z., 2015. SLA in Cloud Computing

Architectures: A Comprehensive Study. Int. J. Grid

Distrib. Comput. 8, 7–32. https://doi.org/10.14257/

ijgdc.2015.8.5.02.

Arabnejad, H., Jamshidi, P., Estrada, G., El Ioini, N., Pahl,

C., 2016. An auto-scaling cloud controller using fuzzy

q-learning-implementation in openstack, in: European

Conference on Service-Oriented and Cloud Computing.

Springer, pp. 152–167.

Banerjee, A., Agrawal, P., Iyengar, N.C. h. S.N., 2013.

Energy Efficiency Model for Cloud Computing. Int. J.

Energy Inf. Commun. 4, 29–42. https://doi.org/10.142

57/ijeic.2013.4.6.04.

Buyya, R., Saurabh Kumar, Rodrigo N. Calheiros (Eds.),

2011. SLA-Oriented Resource Provisioning for Cloud

Computing: Challenges, Architecture, and Solutions.

Hoboken, N. J.

Carlsson, C., Fullér, R., 2013. Probabilistic versus

possibilistic risk assessment models for optimal service

level agreements in grid computing. Inf. Syst. E-Bus.

Manag. 11, 13–28. https://doi.org/10.1007/s10257-

011-0187-z.

Chen, G., Pham, T.T., 2001. Introduction to fuzzy sets,

fuzzy logic, and fuzzy control systems. CRC Press,

Boca Raton, FL.

De Marco, L., Ferrucci, F., Kechadi, M.-T., 2015. SLAFM

- A Service Level Agreement Formal Model for Cloud

Computing: SCITEPRESS - Science and and

Technology Publications, pp. 521–528. https://doi.org/

10.5220/0005451805210528.

Djemame, K., Padgett, J., Gourlay, I., Armstrong, D., 2011.

Brokering of risk-aware service level agreements in

grids. Concurr. Comput. Pract. Exp. 23, 1558–1582.

https://doi.org/10.1002/cpe.1721.

Furht, B., Escalante, A. (Eds.), 2010. Handbook of Cloud

Computing. Springer US, Boston, MA.

Goyal, S., Bawa, S., Singh, B., 2016. Green Service Level

Agreement (GSLA) framework for cloud computing.

Computing 98, 949–963. https://doi.org/10.1007/

s00607-015-0481-6.

Horri, A., Mozafari, M.S., Dastghaibyfard, G., 2014. Novel

resource allocation algorithms to performance and

energy efficiency in cloud computing. J. Supercomput.

69, 1445–1461. https://doi.org/10.1007/s11227-014-

1224-8.

Jamshidi, P., Ahmad, A., Pahl, C., 2014. Autonomic

resource provisioning for cloud-based software.

Presented at the 9th International Symposium on

Software Engineering for Adaptive and Self-Managing

Systems, ACM Press, India, pp. 95–104.

https://doi.org/10.1145/2593929.2593940.

Jamshidi, P., Sharifloo, A., Pahl, C., Arabnejad, H.,

Metzger, A., Estrada, G., 2016. Fuzzy Self-Learning

Controllers for Elasticity Management in Dynamic

Cloud Architectures. Presented at the 12th

International ACM SIGSOFT Conference on Quality of

Software Architectures, IEEE, pp. 70–79.

https://doi.org/10.1109/QoSA.2016.13.

Jamshidi, P., Sharifloo, A.M., Pahl, C., Metzger, A.,

Estrada, G., 2015. Self-Learning Cloud Controllers:

Fuzzy Q-Learning for Knowledge Evolution. Presented

at the International Conference on Cloud and

Autonomic Computing, IEEE, Cambridge, MA, USA,

pp. 208–211. https://doi.org/10.1109/ICCAC.2015.35.

Josyula, V., 2012. Cloud computing: automating the

virtualized data center. Cisco Press, Indianapolis, IN.

Kliazovich, D., Bouvry, P., Khan, S.U., 2013. DENS: data

center energy-efficient network-aware scheduling.

Clust. Comput. 16, 65–75. https://doi.org/10.1007/

s10586-011-0177-4.

Mahmood, Z. (Ed.), 2014. Cloud Computing, Computer

Communications and Networks. Springer International

Publishing, Cham.

Marotta, A., Avallone, S., Kassler, A., 2018. A Joint Power

Efficient Server and Network Consolidation approach

for virtualized data centers. Comput. Netw. 130, 65–80.

https://doi.org/10.1016/j.comnet.2017.11.003.

Moreno, I.S., Xu, J., 2011. Customer-aware resource

overallocation to improve energy efficiency in realtime

cloud computing data centers, in: Service-Oriented

Computing and Applications (SOCA), 2011 IEEE

International Conference On. IEEE, pp. 1–8.

Ragmani, A., El Omri, A., Abghour, N., Moussaid, K.,

Rida, M., 2016. A performed load balancing algorithm

for public Cloud computing using ant colony

optimization, in: 2nd International Conference

(CloudTech 2016). IEEE, Marrakech, Morocco, pp.

221–228.

Teixeira Sá, T., Calheiros, R.N., Gomes, D.G., 2014.

CloudReports: An Extensible Simulation Tool for

Energy-Aware Cloud Computing Environments, in:

Mahmood, Z. (Ed.), Cloud Computing. Springer

International Publishing, Cham, pp. 127–142.

https://doi.org/10.1007/978-3-319-10530-7_6.

Verma, D.C., 2004. Service level agreements on IP

networks. Proc. IEEE 92, 1382–1388. https://doi.org/

10.1109/JPROC.2004.832969.

Wieder, P., Butler, J.M., Theilmann, W., Yahyapour, R.

(Eds.), 2011. Service Level Agreements for Cloud

Computing. Springer New York, New York, NY.

https://doi.org/10.1007/978-1-4614-1614-2.

A Novel Green Service Level Agreement for Cloud Computing using Fuzzy Logic

665