Selecting the Right Cloud Service(s)

Abstract

Cloud computing is a strongly growing business, and more
and more companies are entering the market of cloud services. With a
broader range of such services, the choice between them becomes in-
creasingly difficult and many businesses, which want to replace their
internal IT infrastructure through an external cloud solution, are faced
with a challenging choice. This paper provides a methodology to facili-
tate the selection of the most suitable cloud service containing a tech-
nological suitability analysis as well as a service comparison using eight
well-defined cloud computing characteristics.

Full text

Selecting the Right Cloud Service(s)
Manuel R¨osch
Department of Informatics,
University of Zurich, Switzerland
manuel.roesch@uzh.com
Abstract. Cloud computing is a strongly growing business, and more
and more companies are entering the market of cloud services. With a
broader range of such services, the choice between them becomes in-
creasingly difficult and many businesses, which want to replace their
internal IT infrastructure through an external cloud solution, are faced
with a challenging choice. This paper provides a methodology to facili-
tate the selection of the most suitable cloud service containing a tech-
nological suitability analysis as well as a service comparison using eight
well-defined cloud computing characteristics.
Keywords: Cloud Computing, Cloud Service, Service Selection, Multi-
Criteria Decision Making, Analytic Hierarchy Process, Decision Support,
Technological Suitability Analysis
1 Introduction
According to [3] the usage cloud computing will more than triple from 2014 to
2019, which means that in coming years, many companies will face the deci-
sion between different cloud service offers. Therefore, it is a relevant research
topic to improve and facilitate such a decision process, especially because com-
plex deployment options arose with the introduction of cloud computing. So
far, companies have bought a fixed hard- and software infrastructure over which
they have full control and access for a more or less fixed price. However, with
the introduction of cloud services, different options arose which all underlie dif-
ferent contracts varying in the pricing model, the allowed control over hard- and
software as well as the guaranteed performance. Besides, there are new risks like
a vendor lock-in where a customer cannot easily switch to a competitor’s ser-
vice because he built a certain dependence on the service provider through the
usage of proprietary technology. All these issues result in a challenging decision
between distinct services and providers.
In literature, different tools and techniques can be found that support such a
decision but many of them only compare similar services or service models based
on few attributes like cost or performance [9][11]. This work, in contrast, brings
up a broader perspective, considering various aspects of the decision process.
The remainder of this work is structured as follows. Section 2 gives an
overview of the different types of cloud services, and Section 3 presents essen-
tial cloud service characteristics. The actual cloud service selection process is

2 Selecting the Right Cloud Service(s)
covered in Section 4 presenting a methodology containing five well-defined steps
to select the right cloud service. Finally, Section 5 summarizes this work, draws
conclusions and discusses future work.
2 Definition and Types of Cloud Computing Services
This section starts with the definition of cloud computing to ensure a common
understanding of the term, which is required for the rest of this work. The
definition used for this purpose is from the National Institute of Standards and
Technology (NIST):
“Cloud computing is a model for enabling ubiquitous, convenient, on-demand
network access to a shared pool of configurable computing resources (e.g., net-
works, servers, storage, applications, and services) that can be rapidly provi-
sioned and released with minimal management effort or service provider inter-
action.” [20]
This definition contains all the key aspects of cloud computing and should be
used as an indicator to which kind of services the selection guideline, presented
in this paper, applies.
2.1 Service Models
Besides that definition, the NIST [20] suggests that cloud services can be classi-
fied into three distinct service models:
– Software as a Service (SaaS)
SaaS describes applications, which are stored on the infrastructure of a cloud
service provider and can be accessed by the customer over the Internet from
various devices either through thin client interfaces (e.g. web browsers) or a
program interface. The customer does not have to manage the infrastructure
and the application on which he is working, and the customization is strongly
limited to the options directly provided. [20]
Salesforce [22] and Google Apps [13] are prominent examples of SaaS. [14]
– Platform as a Service (PaaS)
PaaS provides to the customer a managed higher-level software infrastruc-
ture where supported classes of applications can be deployed. The infrastruc-
ture is configured and maintained by the service provider, and it is abstracted
away below the platform so that the user has limited control over it. PaaS
providers usually support only certain specific programming languages, li-
braries, services and tools. [20]
As an example for a PaaS, the Google App Engine [12] can be brought up.
[14]
– Infrastructure as a Service (IaaS)
IaaS offers an infrastructure on which a customer can deploy his software on
virtual machines. Thereby, he has full control over his own virtual machine
and can control and configure it according to his requirements. The virtual

Selecting the Right Cloud Service(s) 3
machines can be rented for the desired time (e.g. only one hour), and the
infrastructure is dynamically scalable to the needs of the customer. [20]
Amazon Elastic Compute Clouds (Amazon EC2) [2] is a popular example of
IaaS. [14]
3 Essential Characteristics of Cloud Services
The selection of the essential cloud service characteristics differs among liter-
ature [8][10][14][17][20][23]. The characteristics used in this work are based on
the Service Measurement Index (SMI) [8] which is developed by the Cloud Ser-
vices Measurement Initiative Consortium (CSMIC) to help organizations mea-
sure cloud-based services based on their requirements. The SMI contains more
than 40 characteristics and because this is too much for the scope of this work,
a selection of the 21 most important SMI characteristics is used. These 21 char-
acteristics are determined by a study from Siegel and Perdue in which several
interviews with CSMIC members and SMI clients were performed. [23] These
findings are also supported by [7] which came to a similar set of attributes.
Figure 1 gives an overview of the proposed essential characteristics, and Sub-
section 3.1 to 3.7 explain them in more detail.
accountabli
ty
ag
il
ity
assurance financial
performan
ce securi
ty & priva
cy usabili
ty
Cloud Service
Characteristics
Provider contract/
SLA verification
Compliance
Ease of doing
business
Provider
certification
Scalability
Portability
Extensibility
On-going
costs
Acquisition
and transition
cost
Accessibil
ity
Learnability
Suitablity
Service response
time
Functionality
Interoperability
Access control &
privilege management
Data privacy and
data loss
Data integrity
Availability
Reliability
Resiliency/
Fault
tolerance
Fig. 1. The essential cloud service characteristics based on the SMI [8] and [23].
3.1 Accountablity
This category contains attributes related to the properties of the provider orga-
nization. Relevant attributes are:

4 Selecting the Right Cloud Service(s)
Provider Contract/Service Level Agreement (SLA) verification: Most
cloud services are subject to a contract and/or SLA. Such a contract respectively
agreement contains important information about the abilities and the attitude
of a provider. Central are the availability and performance guarantees as well as
the compensation if such a guarantee cannot be fulfilled. The question is thereby
if the contract/agreement is adequate to manage the cloud service and also
mitigates the risk of a service failure on an agreeable level. Besides, a contract
can contain important provisions on the rights over the clients’ data, software
licenses, and intellectual property. [8]
Compliance: This attribute is about the provider’s attitude towards his com-
mitments and evaluates if the provider follows the promised standards, processes
and policies. [8]
Ease of doing business: Providers differ in how they interact with their cus-
tomers and because interactions are on a regular basis (e.g. monthly billing
process) it is crucial that these interactions take place in such a manner that it
is easy for the client to do his business. Important are the provider’s business
stability and transparency so that a client can plan in advance and do not face
any surprises as well as the simplicity of the whole billing process. Besides, com-
munication is also a central point. In the best case, the provider can be contacted
by phone, and competent and professional help is provided. [8]
Provider Certification: Certificates guarantee that specific standards and
guidelines are fulfilled by a provider. Because cloud services are compared to
traditional IT infrastructure less transparent regarding the whole techniques
and mechanisms in the background, certificates can be used as a starting point
to evaluate hidden aspects of a service. The client should go through all the
provider’s certificates and assess if the certificate is important and useful con-
cerning his requirements. [8]
3.2 Agility
This category contains attributes related to the user’s possibilities to change di-
rection, strategy or tactics quickly with a particular service. Relevant attributes
are:
Scalability: Especially for cloud services, scalability is a central attribute be-
cause it is one of the sales arguments that resources can be easily scaled ac-
cording to the requirements of the user. There should be either an automatic or
straightforward manual way to do it. [10]

Selecting the Right Cloud Service(s) 5
Portability: There are several scenarios where it is necessary to move a service
from one to another provider (legal reasons, prices or new requirements). To
keep such a service relocation within a certain price and time limit, portability
should be guaranteed. Portability ensures that there is no vendor lock-in, which
is a big issue with several cloud service providers. [8]
Extensibility: Sooner or later, a service are faced with the demand of new
features because business requirements have changed. To satisfy such a demand,
a service should be extensible so that features can be easily added according to
the requirements. [8]
3.3 Assurance
This category contains attributes related to the stability of a service and how
likely it will work as specified. Relevant attributes are:
Availability: The unavailability of a service can cause both, financial damage
and reputation loss. Therefore, it is recommended to spend some effort on evalu-
ating the availability of a service. Besides the SLA, some independent measure-
ment should be considered (if available) or test measurements can be performed.
[10]
Reliability: The attribute reliability is about the duration of error-free service
operation under certain conditions. A high reliability is crucial that a service is
useful for a customer and should be ensured by a service provider. [10]
Resiliency/Fault tolerance: It is frustrating, for an end-user, if the system
often crashes because of minor errors or insignificant hardware faults. Hence, the
system should be resilient towards small issues and the users should be actively
hindered to crash the system. [8]
3.4 Financial
This category contains attributes related to the money spent on the service.
Relevant attributes are:
On-going costs: Cloud service providers’ typical pricing models are based on
the resources needed by the customer. Resources can either be, in the case of
PaaS/IaaS, the used infrastructure (e.g. CPU, RAM or memory) or, in the case
of SaaS, the number of users who use the service. There are also other pricing
schemes and providers often have a price calculator or at least a price table where
a user can estimate the monthly costs. A sophisticated evaluation of the ongoing
expenses should consider more than only the height of the monthly payments.
Relevant are also the payment possibilities (e.g. monthly or yearly, payment
automation) and if the payment rules are transparent, scalable and predictable.
[10]

6 Selecting the Right Cloud Service(s)
Acquisition and transition cost: During the period of setup and deploy-
ment of a service, there can be some initial costs. These costs arise because the
provider charges a payment for the service setup or because there are some ex-
penses during the development, migration and deployment of a service. Besides,
costs may occur due to the necessary restructuring of infrastructure or human
resources. [10] Additionally, hidden or non-obvious costs like the expenses for the
employee training can have a big influence on the total initial costs, considering
that both, the recruitment of trainers and the loss of employee working hours
must be taken into account. [10]
3.5 Performance
This category contains attributes related to the features and functions provided
by the service. Relevant attributes are:
Service response time: This attribute measures the time between a service
request and the availability of its response. A low response time guarantees
afluent interaction with the service what is important for a satisfying user
experience. The geographic location of a cloud service and its servers often has
a big impact, and therefore, a service should be located not too far away from
its users. Independent measurements should be considered (if available) or test
measurements can be performed to get an accurate opinion about this attribute.
[10]
Functionality: Every service provides a certain functionality, and a customer
should examine the range of available functions and features. The more features
a service has, the more likely is it that current and future tasks can be resolved.
Functionality can, depending on the service, include different attributes but it
is to mention that the functionalities concerning agility, assurance, security and
usability are most likely covered by the other categories and should be considered
there. [8]
Interoperability: If the service has interfaces, which can be linked to other
services, it can be extended and the power of different services can be united.
Such an interoperability can increase the total potential of a system and is,
therefore, a relevant attribute. [10]
3.6 Security and Privacy
This category contains attributes related to the effectiveness of a cloud service
provider’s service and data access control. Relevant attributes are:

Selecting the Right Cloud Service(s) 7
Access control/privilege management: A service should contain measures
to ensure confidentiality and data privacy. This can be assured by encrypted
connections and sophisticated access control mechanisms, which includes, for
example, a multi-stage login process. With such measures, it is ensured the data
can only be used and modified by personnel with appropriate privileges. [8][14]
Data privacy and data loss: Data theft and data loss usually result in finan-
cial or reputational damage. Either because the lost work has to be repeated or
because important business secrets are disclosed. Therefore, a service provider
should ensure a high level of data privacy and sophisticated methods against
data loss. [8]
Integrity: Service integrity means that the data is in an accurate and consistent
state over the entire life-cycle, and the user can be confident that it is valid. To
ensure integrity, a service must provide a secure connection establishment to
prevent man-in-the-middle attacks and besides it must be aware of concurrent
data modifications and handle it appropriately. [8][14]
3.7 Usability
This category contains attributes related to the ease of service usage. Relevant
attributes are:
Accessibility: A service should be operable by users with disabilities. Accessi-
bility measures to which degree this is possible. [8] The Web content accessibility
guideline of the World Wide Web Consortium [26] contains important attributes
like perceptibility, operability, understandability and robustness, which can be
used to examine how accessible the service is.
Learnability: To get the acceptance of the users of a service, it is crucial that
a service underlies a steep learning curve. Otherwise, the frustration potential
is high and some user might refuse a service before all the capabilities of the
service are revealed. Besides, it is also a question of money: If the learnability
is poor, the costs for employee training and/or for software developers are much
higher. [8]
Suitablity: The attribute suitability covers the aspect how close the available
features comply with the user’s requirements. Suitability is crucial because the
existence of certain features may decide if a service is useful for a customer or
not. [10]

8 Selecting the Right Cloud Service(s)
4 Cloud Service Selection
In this section, a guideline for selecting the right cloud service is provided using
afive-step selection methodology as illustrated in Figure 2. Each of the five steps
is explained in the Subsection 4.1 to 4.5 in more detail.
Final Selection
Cloud Service
Rating using AHP
Initial Cloud
Service Selection
Suitability
Analysis
Preparation
Fig. 2. The five step cloud service selection process.
4.1 Preparation
Afirst and fundamental step is the preparation. The aim of this step is to
get an overview of cloud technology and the available cloud services. This is
necessary for all the following steps. Important is thereby the understanding of
the definition of cloud computing so that this guideline is not applied to wrong
products or services. Besides, the three service models, as well as the essential
characteristics, should be well understood so that this service selection guideline
can be applied correctly and professionally.
A comprehensive understanding of cloud computing increases the efficiency
and effectiveness of the whole selection process, and therefore, a deeper under-
standing is desirable. The book [24] about cloud computing is recommended for
that purpose as well as [1] and [19] which present and compare distinct cloud
services. There also exist online cloud service directories like Cloudsquare [4] and
Coudxl [5] which can be visited for the exploration of different services.
Furthermore, it is also advised that the websites of cloud service providers
are directly visited, and their offers examined because information in literature
can be outdated [3]. In addition, it is suggested to use Internet search engines
to explore new service options and providers, especially in the area of SaaS.

Selecting the Right Cloud Service(s) 9
4.2 Suitability Analysis
The next step is a suitability analysis where the user has to ensure that the
technology of cloud computing is suitable under given requirements otherwise
it might be better to stop the cloud service selection process and retain the
existing system or change to another alternative. To facilitate the suitability
analysis, this paper provides a checklist containing eight prerequisites for cloud
computing. These eight points, based on the work of [17], are summarized in
Table 1.
Table 1. The eight prerequisites of cloud technology.
Characteristic Prerequisites
1. Elasticity Software architecture supports scaling out or
scaling up to a bigger server is sufficient
2. Communications Bandwidth speed and latency within and be-
tween the cloud are acceptable
3. Processing CPU power as well as memory of cloud ser-
vice instances are sufficient considering the ex-
pected operating load
4. Availability/
dependability
There are services underlying a suitable SLA
or it is possible to reach the demanded avail-
ability by mixing locations and providers
5. Security requirements There is a provider which meets your desired
security requirements (multi-factor authenti-
cation or encrypted data transfer)
6. Data confidentiality
and privacy
Data confidentiality and privacy guarantees
are sufficiently high
7. Access to hardware/
bespoke hardware
The required access to hardware components
or bespoke hardware is provided
8. Regulatory
requirements
The required regulatory requirements are com-
plied
Although the checklist is not exhaustive, many aspects are covered concerning
a suitability analysis and therefore, it can be used as a first assessment whether
or not to use a cloud service and proceed with further analysis.
Point 1-3 in Table 1 cover technical properties of cloud services. Point 1
checks if the deployed service can take advantage of the elasticity of cloud ser-
vices, and 2 and 3 verify that communication and processing capabilities are
sufficient for the service user. If one of these points are not fulfilled it is chal-
lenging to take full advantage of cloud services. [17]
Point 4-6 in Table 1 are related to the risks of cloud service usage. In case
these points are not fully satisfied, the usage of a cloud service is still possible
but brings some risks with it. In a further risk analysis, it should be evaluated if
switching to a cloud infrastructure is worth the risk or if an alternative option
is more reasonable. [17]

10 Selecting the Right Cloud Service(s)
Point 7 and 8 in Table 1 examine crucial prerequisites of cloud services
that are hardly changeable. If either of them does not comply with the user’s
requirements, it is likely that the use of a cloud service is inhibited, and other
alternatives should be taken into account. [17]
4.3 Initial Cloud Service Selection
After having determined cloud technology as suitable, the next step has the
objective to identify a set of possible cloud services. Such a pre-selection is
necessary because the rating of services is demanding and time-consuming and
therefore, it can be applied only to a small number and not for all of them.
First, it should be determined which of the service models are suitable. Table
2 serves as a guideline and summarizes when to use and when not to use the
different service models, and it should be applied to determine, which of them
fit best the requirements.
Table 2. Service model selection key points based on [16].
Service
Model
When to use When not to use
SaaS –undifferentiated vanilla offerings
without competitive advantage
–significant interplay between or-
ganization and the outside world
is needed e.g. newsletter software
–significant need for web or mobile
access e.g. collaboration software
–regular demand spikes
–extremely fast processing of real-
time data is required
–legislation and other regulations
prohibit externally hosted data
–existing on-premise solution ful-
fills all the organization’s needs
PaaS –external parties interact with the
development process
–developers wish to automate test-
ing and deployment services
–highly portable application
–proprietary languages or ap-
proaches impact development
–proprietary language causes ven-
dor lock-in
–customization of the underlying
hardware/software is required
IaaS –irregular/volatile demand spikes
–new organizations without the
capital to invest in hardware
–scaling hardware is problematic
(e.g. fast growing organization)
–to limit capital expenditure
(move to operating expenditure)
–for specific business lines, trial or
temporary infrastructural needs
–data storage and processing out-
sourcing are difficult due to regu-
latory compliance
–highest levels of performance are
required and on-premise or ded-
icated hosted infrastructure has
the capacity to meet the organi-
zation’s needs

Selecting the Right Cloud Service(s) 11
Having found suitable service model(s) the number of available options is
decreased, and the selection facilitated. What follows the service model selection
is the exploration and selection of services. To explore the different services,
there are the lists of distinct cloud service products in [1] and [19] as well as
online cloud service directories like [4] and [5]. Furthermore, it is suggested that
search engines are used to explore the most-recent services that are possibly not
listed in the mentioned literature and service directories.
4.4 Cloud Service Rating using Analytic Hierarchy Process (AHP)
The next step, after the selection of potential cloud services, is to compare and
rank them. This ranking can be interpreted as a multi-criteria decision-making
problem and this work uses the AHP method, which is well-known in literature
[10][11][15][21], to solve this problem.
The AHP method expects for every solution of the problem one or more
criteria which are hierarchically structured so that each criterion has some sub-
criteria respectively a sub-criteria hierarchy. On each level of the criteria hierar-
chy, weights are assigned and normalized so that the sub-criteria weights of each
criterion sum up to one. On the lowest level of the criteria hierarchy, the criteria
are then rated, and these ratings are normalized as well. [21]
After the rating is done, the weights are multiplied all the hierarchy down and
the resulting weight is multiplied with the associated rating. The resultant score
of this calculation is summed up with the score of other criteria for a particular
solution, and this results in a final score that can be compared to the scores of
the other solutions. [21]
In the following, it is explained how the AHP method can be applied to a
cloud service selection problem. The method is explained in three steps (weight-
ing, rating, scoring) using ratings based on [7] for Google Apps (P1) and Office
365 (P2) and fictive weights.
Weighting: The first step of the AHP process is the weighting of the different
SMI categories and the weighting of the attributes of each category according to
their estimated relative importance. The weightings of all category weights, as
well as the attribute weights, has to be normalized. The result of the weighting
can be seen in Table 4 in the first two columns.
Although the categories and attribute hierarchy is well-selected, it can and
should be adapted to the requirements of the user who can also expand it with
more attributes or more hierarchy levels.
Rating: The step that follows the weighting is the rating. This step assigns a
rating to each of the attributes. This work proposes a four-star rating scale where
each attribute gets a certain number of stars assigned. The proposed rating scale
is based on the improved mean opinion score scale of [18] and is presented in
Table 3.

12 Selecting the Right Cloud Service(s)
Table 3. Proposed rating scaled based on [18].
Rating Description
���� Excellent/Perfect
��� Good
�� Ordinary
�Poor
Awful/Useless
After the assignment of a rating to all the attributes of each provider, it is
necessary to normalize them. Although, different normalization techniques can
be applied for that [21], this paper suggests a linear normalization using Equation
1 to normalize attribute ratings.
NormalisedAttributeRatingi=AttributeRatingi
N
n=1 AttributeRatingn
(1)
Scoring: In a next step, the weighted ratings of each attribute are calculated by
multiplying the attribute weighting with the corresponding category weighting
and the attribute rating. Equation 2 and 3 illustrate this calculation, and Equa-
tion 4 shows an exemplary calculation using the weightings and ratings from
Table 4.
ScoreP rovideri=N
n=1(W eig htAttributen
·RatingP rovideri,Attr ibten)(2)




Scorep1
Scorep2
...
Scorepn




T
=



Wa1
Wa2
...
Wan




T



Rp1,a1Rp2,a1... Rpn,a1
Rp1,a2Rp2,a2... Rpn,a2
... ... ... ...
Rp1anRp2an... Rpnan




(3)
Scorep1
Scorep2T
=












0.2·0.4
0.2·0.3
0.2·0.2
0.2·0.1
0.12 ·0.5
0.12 ·0.3
...
0.1·0.3












T











0.429 0.571
0.500 0.500
0.571 0.429
0.500 0.500
0.571 0.429
0.800 0.200
...
0.571 0.429












=0.518
0.482 T
(4)
After the calculation and summation of all the weighted ratings, there is a
resulting score that represents the relative suitability of a provider: The higher
the score, the more suitable the provider is. In case of P1 (Google Apps) and P2
(Office 365) in Table 4, P2 has a higher score and would therefore be considered
as the more suitable solution.

Selecting the Right Cloud Service(s) 13
Table 4. AHP provider rating using ratings based on [7] for the SaaS products Google
Apps (P1) and Office 365 (P2).
Category Attribute Rating Star/Normalized Score
Weights Weights P1 P2 P1 P2
Accountablity 0.20 Provider Contract/
SLA verification 0.4 ��� 0.43 ���� 0.57 0.034 0.046
Compliance 0.3 ��� 0.50 ��� 0.50 0.030 0.030
Ease of doing
business 0.2 ���� 0.57 ��� 0.43 0.023 0.017
Provider
certification 0.1 ���� 0.50 ���� 0.50 0.010 0.010
Agility 0.12 Scalability 0.5 ���� 0.57 ��� 0.43 0.034 0.026
Portability 0.3 ���� 0.80 �0.20 0.029 0.007
Extensibility 0.2 �0.50 �0.50 0.012 0.012
Assurance 0.08 Availability 0.4 ���� 0.50 ���� 0.50 0.016 0.016
Reliability 0.3 ���� 0.57 ��� 0.43 0.014 0.010
Resiliency/
Fault tolerance 0.3 ��� 0.50 ��� 0.50 0.012 0.012
Financial 0.15 On-going costs 0.7 ��� 0.60 �� 0.40 0.063 0.042
Acquisition and
transition cost 0.3 ��� 0.60 �� 0.40 0.027 0.018
Performance 0.10 Service response
time 0.4 ���� 0.57 ��� 0.43 0.023 0.017
Functionality 0.3 ��� 0.43 ���� 0.57 0.013 0.017
Interoperability 0.3 �� 0.50 �� 0.50 0.015 0.015
Security and
Privacy 0.25
Access control/
privilege
management
0.5 ���� 0.50 ���� 0.50 0.063 0.063
Data privacy and
data loss 0.3 ��� 0.43 ���� 0.57 0.032 0.043
Data integrity 0.2 ��� 0.43 ���� 0.57 0.021 0.029
Usability 0.10 Accessibility 0.4 ��� 0.43 ���� 0.57 0.017 0.023
Learnability 0.3 ��� 0.43 ���� 0.57 0.013 0.017
Suitablity 0.3 ���� 0.57 ��� 0.43 0.017 0.013
Total Score 0.518 0.482

14 Selecting the Right Cloud Service(s)
4.5 Final Selection
In the final selection step, the two or three best-ranked options are further in-
vestigated. This paper suggests a test installation as well as the usage of cloud
computing modeling and simulation tools like CloudSim [6] or SPECI [25] to per-
form further analysis and comparison. Besides, an exhaustive Internet research
about the highest ranked solutions may bring some important insights, which
can lead to an elaborated final selection.
5 Conclusion and Future Work
The market of cloud computing is strongly growing, and more and more busi-
nesses consider a change to cloud technology [3]. The resulting selection of a
suitable cloud service satisfying the requirements best is a challenging task.
This paper provides a comprehensive methodology for determining (1) whether
cloud technology is applicable to a particular customer and (2) which service
fits most to the customer’s requirements. For the latter determination, 21 well-
defined cloud service characteristics based on the SMI [23] and [8] are compared,
weighted and ranked using the AHP method. This structured method applied to
the well-selected service characteristics and attributes ensure a comprehensible
and transparent way to make an elaborated and unbiased decision.
For future work, the proposed methodology must be tested in a real-world
scenario, and possible shortcomings must be evaluated. In addition, the usage of
the proposed selection guideline could be facilitated by building a web interface
that guides the user through all the five steps and automatically calculates the
scores of the AHP provider comparison. Besides, the process of exploring cloud
services must be facilitated. This could be done by building a comprehensive
online cloud service directory with elaborated search and filtering possibilities
that allows to compare providers and their offers and there should be also some
direct information about the service characteristics so that the comparison and
rating are facilitated.
References
1. Alam, M. I., Pandey, M., Rautaray, S. S.: A Comprehensive Survey on Cloud Com-
puting. International Journal of Information Technology and Computer Science
(IJITCS), p. 68 (2015)
2. Amazon EC2. https://aws.amazon.com/de/ec2/
3. Cisco: Global Cloud Index: Forecast and Methodology, 2014–2019 White Paper
(2015)
4. CloudHarmony: Provider Directory. https://cloudharmony.com/cloudsquare
5. CloudXL: An Organized List of Cloud Computing and Software as a Service
Providers. http://www.cloudxl.com/
6. The Cloud Computing and Distributed Systems (CLOUDS) Laboratory, University
of Melbourne: CloudSim - A Framework for Modeling and Simulation of Cloud
Computing Infrastructures and Service. http://www.cloudbus.org/cloudsim/

Selecting the Right Cloud Service(s) 15
7. Costa, P., Santos, J. P., da Silva, M. M.: Evaluation criteria for cloud services. In
Cloud Computing (CLOUD), 2013 IEEE Sixth International Conference on Cloud
Computing (pp. 598-605). IEEE (2013)
8. CISMIC: Service Measurement Index Framework Version 2.1. Carnegie Mellon Uni-
versity Silicon Valley (2014)
9. Frey, S., Fittkau, F., Hasselbring, W. Search-based genetic optimization for de-
ployment and reconfiguration of software in the cloud. In Proceedings of the 2013
International Conference on Software Engineering, pp. 512-521. IEEE (2013)
10. Garg, S. K., Versteeg, S., Buyya, R.: SMICloud: A framework for comparing and
ranking cloud services. In Utility and Cloud Computing (UCC), 2011 Fourth IEEE
International Conference on Cloud Computing, pp. 210-218, IEEE (2011)
11. Godse, M., Mulik, S.: An approach for selecting software-as-a-service (SaaS) prod-
uct. In Cloud Computing, 2009. IEEE International Conference on Cloud Comput-
ing, pp. 155-158, IEEE (2009)
12. Google App Engine. https://cloud.google.com/appengine/
13. Google Apps. https://apps.google.com/
14. Hoefer, C. N., Karagiannis, G.: Cloud computing services: taxonomy and compar-
ison. Journal of Internet Services and Applications, pp. 81-94. (2011)
15. Hussain, O. K., Hussain, F. K.: Iaas cloud selection using MCDM methods. In
e-Business Engineering (ICEBE), 2012 IEEE Ninth International Conference on
e-Business Engineering, pp. 246-251. IEEE (2012)
16. Kepes, B. Understanding the Cloud Computing Stack. Rackspace (2011).
17. Kha jeh-Hosseini, A., Greenwood, D., Smith, J. W., Sommerville, I.: The cloud
adoption toolkit: supporting cloud adoption decisions in the enterprise. Software:
Practice and Experience 42, pp. 447-460. (2012)
18. Kilkki, K., An Introduction to Communications Ecosystems, In Measuring service
quality, pp. 72-77. CreateSpace (2012)
19. Laszewski, G., Diaz, J., Wang, F., Fox, G. C.: Comparison of multiple cloud frame-
works. In Cloud Computing (CLOUD), 2012 IEEE 5th International Conference on
Cloud Computing, pp. 734-741. IEEE (2012)
20. Mell, P., Grance, T.: The NIST definition of cloud computing. (2011)
21. Rehman, Z. U., Hussain, F. K., Hussain, O. K.: Towards multi-criteria cloud ser-
vice selection. In Innovative Mobile and Internet Services in Ubiquitous Computing
(IMIS), 2011 Fifth International Conference on Cloud Computing, pp. 44-48. IEEE
(2011)
22. Salesforce. http://www.salesforce.com
23. Siegel, J., Perdue, J.: Cloud services measures for global use: the Service Measure-
ment Index (SMI). In SRII Global Conference 2012, pp. 411-415, IEEE (2012)
24. Srinivasan, S.: Cloud Computing Basics. Springer, New York (2014)
25. Sriram, I. L.: Simulation Program for Elastic Cloud Infrastructures. http://www.
speci.org/
26. World Wide Web Consortium: Web content accessibility guidelines (WCAG) 2.0.
(2008)