A Characterization of Cloud Computing Adoption based on Literature

Evidence

Antonio Carlos Marcelino de Paula

, Glauco de Figueiredo Carneiro

and Rita Suzana P. Maciel

Salvador University - UNIFACS, Salvador, Bahia, Brazil

Federal University of Bahia - UFBA, Salvador, Bahia, Brazil

Keywords:

Cloud Computing, Legacy Systems, Cloud Migration, Studies Report.

Abstract:

Context:The cloud computing paradigm has received increasing attention because of its claimed ﬁnancial and

functional beneﬁts. This paradigm is based on a customizable and resourceful platform to deploy software.

A number of competing providers can support organizations to access computing services without owning

the corresponding infrastructure. However, the migration of information systems and the adoption of this

paradigm is not a trivial task. For this reason, evidence from the literature reporting and analyzing experiences

in this migration should be widely disseminated and organized to be used by companies and by the research

community. Goal:Characterize main strategies and methodologies reported in the literature to describe and

analyze the adoption and migration to cloud computing Method: The characterization followed a four-phase

approach having as a start point the selection of studies published in conferences and journals. Results: Data

gathered from these studies reveal a tendency for companies to choose the public deployment model, the IaaS

service model, the amazon platform, and how the most important characteristics in the cloud adoption decision

are cost, performance, and security and privacy. Conclusion: Due to the variety of strategies, approaches and

tools reported in the primary studies, it is expected that the results in this characterization study would help in

establishing knowledge on how the companies should adopt and migrate to the cloud. These ﬁndings can be a

useful reference to develop guidelines for an effective use of cloud computing.

1 INTRODUCTION

Cloud computing (CC) has increased its adoption by

enterprises in an attempt to include agile, ﬂexible and

provident practices in their TI infrastructure. This

platform has also changed the way information sys-

tem users deal and perceive computing (Weiss, 2007).

The possibility to remotely use hardware and software

resources, as well as the expectation of economies of

scale are main reasons that drive the shift for migrat-

ing existing core business applications to the cloud

(Hashmi et al., 2011). Considering the availability of

a plethora of cloud services and providers (Stieninger

et al., 2014), many companies have engaged to trans-

fer their business processes mechanisms to the cloud

platform (Sadighi, 2014).

This scenario has created opportunities for en-

terprises that have manifested perceived inclination

toward cloud computing and the beneﬁts reaped by

them such as low start-up cost, pay only for utilized

services, up-to-date resources, features, and rapid de-

ployment (Buyya et al., 2009)(Li et al., 2013). On

the other hand, moving to the cloud means giving up

incumbent information systems practices and facing

the initial perception of losing control of data that

in a previous scenario were stored in local servers

(Lee et al., 2013). The paper published in (Hurtaud

and de la Vaissire, 2011) considers that consolidat-

ing huge amounts of data within large public clouds is

also perceived as creating a massive point of failure in

the event of a communication breakdown (impairing

data availability) or espionage activities such as the

recent PRISM programme revelations (a clandestine

mass electronic surveillance data mining programme

created by the NSA - US National Security Agency).

Therefore, it is important to gain an understand-

ing of not only opportunities but also the challenges

regarding the migration and adoption as well as the

reasoning of an attractive cost-beneﬁt relationship

and the selection of service providers that best ﬁt

the stalkeholders needs (Li et al., 2012a) (Li et al.,

2012b). Previous studies have investigated the adop-

tion and acceptance of public cloud services at both

the individual and organizational levels (Chang and

Paula, A., Carneiro, G. and Maciel, R.

A Characterization of Cloud Computing Adoption based on Literature Evidence.

DOI: 10.5220/0006264600530063

In Proceedings of the 19th International Conference on Enterprise Information Systems (ICEIS 2017) - Volume 1, pages 53-63

ISBN: 978-989-758-247-9

Hsu, 2016). From the organizational perspective, sev-

eral studies have identiﬁed the factors that affect cloud

computing adoption,i.e., relative advantage, complex-

ity, compatibility for enterprise (reference omitted

due to the blinded review).

This paper presents a characterization study of

strategies and methodologies reported in the literature

to describe and evaluate the adoption and migration to

cloud computing study to identify evidence from the

literature related to strategies used to conduct studies

focusing on adoption and migration to the cloud com-

puting. The rest of this paper is organized as follows:

Section 2 outlines the research methodology; Section

4 analyzes the selected studies; Section 5 presents and

discusses the results of this characterization. Section

6 presents the threats to validity of the characteriza-

tion. The concluding remarks as well as limitations

and scope for future research have been discussed in

Section 7.

2 METHODOLOGY

This work considered studies selected following the

phases described in Figure 1 to select the studies for

the characterization. The research questions are pre-

sented in Table 1.

2.1 Systematic Literature Review (SLR)

In contrast to a non-structured review process, a Sys-

tematic Literature Review (SLR) (Brereton et al.,

2007) and (Kitchenham and Charters, 2007) reduces

bias and follows a precise and rigorous sequence of

methodological steps to research literature. SLR rely

on well-deﬁned and evaluated review protocols to ex-

tract, analyze, and document results as the stages con-

veyed in Figure 2. This section describes the method-

ology applied for the phases of planning, conducting

and reporting the review.

We aimed to answers the following questions by

conducting a methodological review of existing re-

search:

RQ1. Which strategies are used by companies to

adopt and migrate to the cloud computing? Identify-

ing goals, proposals and motivations for the adoption

of CC, help organizations to better characterize their

needs and therefore provide conditions to a successful

migration. RQ2. Which factors companies consider

to assess the cost-beneﬁt relationship of adoption and

migration to the cloud computing? The knowledge of

the costs and beneﬁts of migration to the CC can be

used as a support for its planning and reference for

other companies. RQ3. How companies select cloud

computing service providers according to their needs

and proﬁle? The knowledge of successful strategies

and problems raised by inappropriate selection of CC

providers allow organizations to be more conﬁdent to

identify providers that best ﬁt their needs.

Publications Time Frame. We conducted a SLR in

journals and conferences papers from January 2005

to June 2016. In a ﬁrst version of this study, we per-

formed the search from January 2005 to June 2015

and in this new version we extended it to June 2016.

The relevant studies selected in the previous step

was the start point for this characterization study

whose planning is described in the next section.

3 PLANNING THE

CHARACTERIZATION

Identify the Needs for a Characterization Study.

To the best of our knowledge, there is no previous

study characterizing strategies, methodologies, ap-

proaches applied in the literature to describe and eval-

uate how companies adopt and migrate to cloud com-

puting. These results have the goal to understand how

companies have adopted and migrated to the cloud,

and how this experiences have been reported in the

literature. This is a requirement to develop guidelines

for an effective use of cloud computing, specially for

newcomers.

Specifying the Research Question. Considering this

context, we have focused on the following research

question: What are the main strategies and prefer-

ences adopted by studies reported in the literature ad-

dressing issues related to the migration to cloud com-

puting? We derived this research question in three

Speciﬁc Research Questions (SRQ) as follows and

presented in Table 1.

Table 1: Speciﬁc Research Questions (SRQ).

SRQ1

How can the studies from the selected

papers be organized to support the

identiﬁcation of main strategies?

SRQ2

How can we classify elements and

strategies adopted in the studies from

the selected papers?

SRQ3

What are the main evidence of

tendencies and preferences in the

planning and execution of the studies

from the selected papers?

Phase 1: Applying the Search String. The applied

search string is presented as follows:

”case study” or ”simulation” or

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

Figure 1: Study Characterization Phases.

Figure 2: Stages of the Study Selection Process containing the Studies included in the extended Version of this SLR.

”benchmarking” or ”experimental results”

or ”experience report” or ”empirical study”

The result of this phase was a list of 39 papers as a

subset of papers from the SLR (reference omitted due

to the blinded review) containing the aforementioned

string.

Phase 2: Selecting the Papers. The 39 selected pa-

pers from the previous phase were analyzed to iden-

tify evidence of strategies adopted in the studies re-

lated to the migration of legacy systems to the cloud.

Papers that do not described their own studies were

discarded. The result of this phase was a list of 19

papers.

Phase 3: Building the Repository. In this phase we

created a repository of 19 selected studies with cor-

responding data. The decision of which type of ele-

ments should be included in the repository was based

on their relevance in the context of migration to the

cloud. We considered as relevant the elements that

were mentioned at least once in the selected stud-

ies. In this case, elements more cited are more rel-

evant in the set of the selected studies. To provide

an intuitive view of how these elements were orga-

nized and to facilitate the identiﬁcation of the strate-

gies used in the studies, we decided to organize them

in a mental model as conveyed in Figure 3. The

nodes are numbered to identify the elements in the

structure. The node 1 represents cloud deployment

models: (1.1) Public, (1.2) Private and (1.3) Hybrid.

The node 2 represents the three service models: (2.1)

SaaS, (2.2) PaaS and (2.3) IaaS. The node 3 repre-

sents possibilities of cloud platforms identiﬁed in the

studies: (3.1) Amazon, (3.2) Azure, (3.3) Coresuite,

(3.4) Force.com, (3.5) Google, (3.6) Openstack, (3.7)

Rackspace and (3.8) Salesforce.com. The node 4 rep-

resents evaluation issues considered in the studies:

(4.1) Agility, (4.2) Auditability, (4.3) Accountability,

(4.4) Cost, (4.5) Performance, (4.6) Scalability, (4.7)

Elasticity, (4.8) Effort, (4.9) Flexibility, (4.10) Qual-

ity Assurance, (4.11) Governability, (4.12) Infrastruc-

ture, (4.13) Interoperability, (4.14) Timeframe, (4.15)

Business Popularity, (4.16) Information Technology

Skilled Staff, (4.17) Security and Privacy, (4.18) Busi-

ness Size and (4.19) Usability. The elements of

node 4.12 represent infrastructure resources and cor-

respond to the following elements: (4.12.1) Storage,

(4.12.2) CPU, (4.12.3) Memory and (4.12.4) Net-

work. The node 5 represents the types of studies

identiﬁed in the selected papers as follows (5.1) Fea-

sibility Study and (5.2) Experience Report. And the

node 6 is related to analysis method: (6.1) Difﬁculties

Analysis, (6.2) Proposed Solution and (6.3) Compar-

ison of Results.

A Characterization of Cloud Computing Adoption based on Literature Evidence

Figure 3: Findings from the Selected Studies.

From the metamodel conveyed in Figure 3, it was

possible to build the repository represented in Table 2

comprised by 19 studies. This repository is the deliv-

erable of Phase 3.

Table 2 represents data related to two speciﬁc re-

search questions: SRQ1 can be mapped to the col-

umn ”Studies” that contains the identiﬁcation of the

selected studies for the characterization. The strategy

and phases to select the papers were described in the

paragraphs following Figure 1. The SRQ2 can be ad-

dressed in the identiﬁcation of elements and strate-

gies in the studies. These elements and strategies

were classiﬁed and presented in the columns as De-

ployment Models, Service Models, Platforms, Char-

acteristics, Types of Study and Analysis Method. Re-

garding the speciﬁc research question (SRQ3), the ev-

idence identiﬁed in the studies will be discussed in the

next section.

Phase 4: Repository Analysis. The strategy to ana-

lyze the repository consisted in identifying the inﬂu-

ence of the selected issues presented in Figure 3 and

according to the levels of abstraction prestend in Fig-

ure 4: Deployment Model, Service Models, Platform

and Characteristics. In the level of Service Model, a

new sub-level was created to contain the following el-

ements Software as a Service (SaaS), Platform as a

Service (PaaS) and Infrastructure as a Service (IaaS).

In the characteristics level, a new sub-level of infras-

tructure characteristics was created to contain the fol-

lowing elements: Memory, Processor and Storage.

In addition, two other levels comprise the repository:

Type of Study and Analysis Method. These two lat-

ter levels are considered as transverse elements when

compared to the other ones previously listed.

Adjusting the Study Characterization Phases. Af-

ter performing the phases presented in Figure 1, we

identiﬁed the opportunity to adjust it by the inclu-

sion of two new phases ”Create the Mental Model”,

represented in Figure 3 and ”Design the Abstraction

Level”, represented in Figure 4. The phase ”Create

the Mental Model” was included to enable the cre-

ation of a visual representation of a hierarchy of the

elements identiﬁed in the analysis. And the ”De-

sign the Abstraction Level” is a visual representation

to organize the identiﬁed elements according to their

groups in the cloud domain. As can be viewed in Fig-

ure 5.

4 DESCRIBING THE STUDIES

In this section, we describe the main characteristics

of the 19 selected studies according to the level of ab-

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

Table 2: Study Repository to Answer SRQ1.

Studies

Deployment

Models (1)

Service

Models (2)

Platform

(3)

Characteristics

(4)

Type of

Studies (5)

Analysis

Method (6)

S2 Public 2.3 3.1 4.4 5.1 6.2

S8 Private 2.3 3.6 NM 5.1 6.3

S9 Public 2.1 3.5 4.4 and 4.8 5.1 6.2

S11 Public 2.3 3.2 4.5 5.1 6.3

S13 Public 2.3 3.1 4.4, 4.12.2 and 4.12.3 5.2 6.2

S16 Public 2.2 and 2.3 3.5 4.4 5.1 6.1

S25 Private 2.3 NM 4.17 5.2 6.1

S31 Public 2.1 and 2.2 3.3, 3.4 and 3.8 4.6 e 4.9 5.2 6.3

S32 Public 2.2 NM

4.4, 4.9, 4.14,

4.15, 4.16, 4.17 and 4.18

5.1 6.1

S40 Public 2.3 3.1

4.4, 4.12.1, 4.12.2,

4.12.3 and 4.12.4

5.2 6.3

S44 Public 2.2 3.2 4.5 5.2 6.3

S45 Hybrid 2.3 3.1 4.4 5.2 6.3

S46 Private 2.1 NM 4.17 5.2 6.2

S49 NM 2.1 NM 4.2, 4.11, 4.13 and 4.17 5.2 6.2

S50 Private 2.2 NM 4.4 and 4.5 5.1 6.3

S54 Public 2.3 3.1 4.4 and 4.5 5.2 6.3

S55 Public 2.2 and 2.3 3.1, 3.2, 3.5 and 3.7 4.4 and 4.5 5.2 6.2

S56 Public 2.3 3.1 and 3.5

4.4, 4.12.1, 4.12.2,

4.12.3 and 4.12.4

5.2 6.3

S57 Public 2.3 3.1, 3.2 and 3.7

4.1, 4.3, 4.4,

4.5, 4.10, 4.17,

4.19, 4.7, 4.12.1,

4.12.2 and 4.12.3

5.2 6.2

NM = Not Mentioned.

Figure 4: Level of Abstraction and Hierarchy of Elements

in the Studies.

straction and hierarchy of elements in the studies pre-

sented in in Figure 4. based on the strategy presented

before. We scrutinized the papers reporting the stud-

ies to identify relevant issues for analysis numbered

according to the items presented in Figure 3.

Study S2: Cost Modeling Tool Evaluation.

Goal: Evaluate a cost modeling tool called Cloud

Adoption Toolkit through the comparison of three cost

options: server acquisition, rental of equivalent in-

frastructure in the cloud, use of elasticity of demand

in the cloud. In the S2 study, the authors evalu-

ated the migration to a public cloud (1.1) using the

IaaS service model (2.3). They also used the Ama-

zon provider web services (3.1) to evaluate migration

costs (4.4) using the cost modeling tool (6) as part of

the Cloud Adoption Toolkit and hence the migration

feasibility (5.1) of speciﬁc services from the St An-

drews University to the cloud.

Study S8: Feasibility of Web Service Migra-

tion. Goal: Evaluate the feasibility (5.1) to migrate

a web service solution to the cloud. For this end, the

S8 study adopted a private cloud (1.2) with a IaaS

(2.3) as a service model. The authors informed that

OpenStack (3.6) was the chosen platform to migrate

two server to the private cloud. The ﬁrst server hosts

a web service while the second hosts a database ser-

vice. The servers were replaced by two virtual ma-

chines instances. The authors did not mentioned the

characteristics as expected in the model presented in

Figure 3. The study presented a feasibility study (5.1)

to compare (6) the results.

A Characterization of Cloud Computing Adoption based on Literature Evidence

Figure 5: Adjusted Version of the Characterization Phases.

Study S9: Evaluation of ARTIST Project Tool.

Goal: The S9 study reports a feasibility study of three

tools aimed at evaluating the maturity, technical fea-

sibility (5.1) and business feasibility (5) in the context

of ARTIST project. The study describes the migration

of a Java application called PetStore to the following

scenario: public cloud (1.1) with the Google App En-

gine (3.5) as platform to evaluate cost (4.4) and effort

(4.8) to migrate the PetStore to the cloud application

in such a way that its users can access the application

as a SaaS (2.2).

Study S11: Feasibility of FTP Server Migra-

tion. Goal: The S11 study reports a feasibility evalu-

ation (5) of a FTP server migration to the cloud with

corresponding advantages and challenges of this mi-

gration. For this end, a FTP server was conﬁgured

in the public deployment model (1.1) using IaaS (2.3)

through the Windows Azure provider (3.2) with elas-

ticity resources (4.7). A performance (4.5) bench-

marking (6.3) was performed aiming at identifying

potential beneﬁts and problems related with the mi-

gration of legacy systems to the cloud.

Study S13: Desktop-to-Cloud-Migration

(D2CM) Tool Evaluation. Goal: The S13 study

reports the evaluation of the D2CM Tool. The goal

is to investigate the performance (4.5) and usability

(4.19) of the tool. In the study, the authors adopted

the public deployment model (1.1), the IaaS service

model (2.3) and the tool D2CM to migrate an envi-

ronment to the Amazon EC2 (3.1). The D2CM tool

integrate a set of software libraries to support both the

migration and the management of experiments in the

cloud. The focus is the evaluation of the processor

(4.12.2), memory (4.12.3) and cost (4.4).

Study S16: Feasibility Study and Difﬁcult

Analysis. Goal: The study focus on the evaluation

of the migration of three legacy systems from the

British Telecom to the Google App Engine (3.5). The

study uses a compatibility checklist to guide the cost

estimation (4.4) to migrate to the public deployment

model (1.1) through the use of a IaaS (2.3) and PaaS

(2.2) service models. In the case of the IaaS, the cost

consisted basically in the management of resources

without the need to change the code. However, when

compared with the PaaS scenario, the cost is higher.

The authors discussed the changes needed to migrate

the three legacy systems to the Google App Engine

(3.5) and the corresponding effort (4.8) and the feasi-

bility (5.1) of the migration.

Study S31: Feasibility Study Goal: Perform a

Feasibility Study (5.1) regarding the Migration of two

legacy systems to the cloud. The authors compared

the results (6.3) of the migration of Enterprise Re-

source Planning (ERP) systems of two companies to

the public cloud (1.1). The ﬁrst migration used both

the SaaS (2.1) and PaaS (2.2) service models com-

bined with the platforms Salesforce.com (3.8) and

Force.com (3.4). On the other hand, the second mi-

gration used the PaaS (2.2) service model aimed at

improving issues such as ﬂexibility (4.9) and scala-

bility (4.6) provided by the features of PaaS.

Study S32: Feasibility Analysis of a Migrating

an Account System to the Cloud. Goal: Present an

experience report regarding the selection of the effec-

tive service model for an accounting system. The fea-

sibility analysis considered the following issues in the

migration decision: public deployment model (1.1),

the PaaS service model (2.2), ﬂexibility (4.9), cost

(4.4), business popularity (4.15), business size (4.18),

security and privacy (4.17), timeframe (4.14), skill

level of the IT staff (4.16). The authors did not men-

tioned the name of the selected platform (3).

Study S40: Selection of an Amazon EC2 In-

stance Conﬁguration. Goal: Identify the best Ama-

zon EC2 (3.1) Instance Conﬁguration in a public

cloud (1.1) using the IaaS (2.3) service model. The

study used a Feature Model to analyze issues related

to infrastructure (4.12) and cost (4.4) to compare a set

of instances conﬁguration to identify the one that best

ﬁt a speciﬁc scenario. The main contributions of the

study is the list of the characteristics as well as con-

ﬁguration options of EC2 together with guidelines for

the conﬁguration of an EC2 instance using the Ama-

zon EC2.

Study S44: Comparing the Performance of a

Legacy System with Its Corresponding Version in

the Cloud. Goal: Collect evidence of the perfor-

mance of a legacy system after its migration to the

cloud and compare with the previous legacy scenario.

The legacy system executed in a .Net platform and in

the cloud used the Microsoft Azure (3.2). The study

concluded that the performance (4.5) does not dete-

riorate when migrating to the cloud. Moreover, the

study presents potential advantages of the PaaS (2.2)

service model in a public cloud (1.1).

Study S45: A Game-theoretic Approach to the

Financial Beneﬁts of Infrastructure-as-a-service.

Goal: Compare the issues of cost (4.4) of an owned

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

data center with the cost of using the platform Ama-

zon EC2 (3.1). The study considered costs per year

for an owned data center as an investment cost amor-

tization and running costs. Investments were con-

sidered as acquisition costs for server and network

hardware together with operation system licenses (3

years write-off) as well as infrastructure and building

costs (15 years write-off). Running costs were con-

sidered as maintenance, power, administration, and

data transfer. The study discussed issues related to ef-

fects of hybrid clouds, reserved instances, economies

of scale and market form, availability risk.

Study S46: Service Provider Selection. Goal:

Select a Service Provider based on security and pri-

vacy (4.17) issues. The scenario is comprised of a

company that aims at providing their point of sale ser-

vices through a SaaS (2.1) in a private cloud (1.1).

Due to conﬁdentiality reasons, the paper does not list

the providers (3) analyzed in the study.

Study S49: Service Provider Selection. Goal:

Evaluate the effectiveness of the Fusion and Aggrega-

tion for Geospatial Information (FAGI) framework to

select service providers. The study discussed the sig-

niﬁcance and ramiﬁcations of a structured selection

of a Cloud Service Provider (CSP) in achieving the re-

quired assurance level based on an organization’s spe-

ciﬁc security posture. The following issues were con-

sidered in the provider selection: security (4.17), au-

ditability (4.2), portability (4.20), governability (4.11)

and interoperability (4.13).

Study S50: Migrating Legacy Applications to

the Cloud. Goal: this case study documents the mi-

gration of a text-mining application, acting as a proxy

for any legacy application, through a set of stages pro-

gressing towards deployment in a cloud environment.

The Private (1.1) deployment model and PaaS (2.2)

Service Model were used. The Platform (3) was not

mentioned. Performance (4.5) and cost (4.4) were

also evaluated in the study.

Study S54: Benchmark of Cloud Computing

Environments. Goal: Conduct a survey on a se-

lection of Cloud providers, and propose a taxonomy

of eight important Cloud computing elements cover-

ing service models (2), resource deployment, hard-

ware/infrastructure (4.12), runtime tuning , business

model (4.11), middleware, and performance (4.5).

Study S55: Evaluation of the CloudCmp Tool.

Goal: Compare the performance (4.5) and cost (4.4)

of AWS (3.1), Microsoft Azure 3.2), Google App

Engine (3.5) and Rackspace (3.7) cloud providers.

CloudCmp measures the elastic computing (4.7),

persistent storage (4.12.1), and networking services

(4.12.4) offered by a cloud along metrics that directly

reﬂect their impact on the performance of customer

applications. For this end, the study analyzed the de-

ployment of three simple applications on the public

cloud (1.1) using the IaaS (2.3) and PaaS (2.2) service

models to check whether the benchmark (6.3) results

from CloudCmp are consistent with the performance

experienced by real applications. In this case, the

study validated the conjecture that CloudCmps results

can be used by customers to choose cloud providers

in lieu of porting, deploying, and measuring their ap-

plications on each cloud. The applications included

a storage intensive e-commerce website, a computa-

tion intensive application for DNA alignment, and a

latency sensitive website that serves static objects.

S56 Study: Evaluate the Effectiveness of

Google Compute Engine (GCE) and Amazon EC2

to Deploy Scientiﬁc Applications. Goal: Use the

Cloud Evaluation Experiment Methodology (CEEM)

to benchmark GCE and compare it with Amazon

EC2, to help understand the elementary capability of

GCE for dealing with scientiﬁc problems.The exper-

imental results and analyses showed both potential

advantages of, and possible threats to applying GCE

to scientiﬁc computing. For example, compared to

Amazons EC2 service, GCE may better suit appli-

cations that require frequent disk operations, while

it may not be ready yet for single VM-based paral-

lel computing. Based on the fundamental evaluation

results, suitable GCE environments can be further es-

tablished for studies focusing on real science prob-

lems.

Study S57: Evaluation of the SMICloud

Framework. Goal: Evaluate the effectiveness of

the Service Measurement Index Cloud (SMICloud)

Framework to rank Cloud services based on QoS re-

quirements. The SMICloud framework provides fea-

tures such as service selection based on QoS require-

ments and ranking of services based on previous user

experiences and performance of services. SMICloud

uses the public cloud (1.1) and the IaaS service model

(2.3) to rank the following providers: Amazon (3.1),

Microsoft Azure (3.2) and Rackspace (3.7). The fol-

lowing characteristics were evaluated: Accountability

(4.3), Agility (4.1), Quality Assurance (4.10), Cost

(4.4), Performance (4.5), Security and Privacy (4.17),

Usability (4.19), Elasticity (4.7) and Infrastructure is-

sues Central Processing Unit (CPU) (4.12.2), Mem-

ory (4.12.3) and Storage (4.12.1).

5 RESULTS AND DISCUSSIONS

In this section, we followed the instructions described

in the phase Characterization the Studies to answer

the Speciﬁc Research Questions SRQ1, SRQ2 and

A Characterization of Cloud Computing Adoption based on Literature Evidence

SRQ3.

We followed the instructions described in ”Phases

for the Studies Characterization” in section 3 to select

the studies to answer the Speciﬁc Research Question

1 (SRQ1). The selected studies were then the refer-

ence to build a repository as presented in Table 2 to

support the identiﬁcation of elements and strategies

used in studies focusing on cloud computing adop-

tion. To answer SRQ2, we analyzed the data avail-

able in the repository (Table 2) to identify elements

and strategies adopted in the selected studies. From

this point, we were able to organize the data in four

levels: deployment models, service model, platform

and characteristics.

Transversely to these four levels, we included an-

other two perspectives: type of studies and analysis

method. To deal with SRQ3, to identify evidence of

patterns and tendencies in the planning and execu-

tion of the selected studies, we analyzed separately

each level previously described considering both the

repository and the summary of each study presented

in the previous section. In the following paragraphs,

we present the results and discuss them.

Deployment Model Analysis. The deployment

model scenario obtained from the selected studies is

conveyed in Figure 6. As can be observed, there

is a tendency for the use of the public deployment

model in the studies, which corresponds to 68% of

the 19 studies. This can be explained by the avail-

ability of the public providers and the myriad of in-

stances types and proﬁles with corresponding pricing

they offer which can meet different study needs and

purposes.

Figure 6: Deployment Models Identiﬁed.

Service Model Analysis. Similarly, the service

model scenario obtained from the selected studies in-

dicated that 68% of the 19 studies use the IaaS service

model. The Figure 7 illustrates this tendency. This

can be explained by the relatively deployment easy

of IaaS (2.3) when compared to PaaS (2.2) service

model. In the ﬁrst case, the deployment can be exe-

cuted through the use of a virtual machine in the cloud

with the full stack of the legacy system to be analyzed

in the study. In the case of PaaS service model, there

is the need to conﬁgure the application in the cloud

using the original platform resources such as operat-

ing systems, databases and drivers.

Figure 7: Service Models Identiﬁed in the Studies.

Platform Analysis. The Figure 8 indicates that 42%

of the studies used the Amazon EC2 platform (3.1). It

should be mentioned that ﬁve studies did not informed

the service provider used in their respective studies.

We could not ﬁnd the reasons for the preference for

the Amazon EC2. However, there is the possibility

that this preference may be the result of the ﬂexibility

of pricing plans for the provider available instances.

Figure 8: Platforms Identiﬁed in the Studies.

Characteristics Analysis. As can be seen in Table

3 and Figure 9, the characteristics that stood out in

the selected studies were: costs (4.4), performance

(4.5) and security/privacy (4.17). From the number

of occurrences, it is possible to conclude that cost is

by far the most inﬂuential characteristic in the cloud

computing adoption. Companies that adopt cloud

computing are willing to pay for resources that can

be allocated in a pay-as-you-go fashion. This can

lead to representative overall cost reduction as a re-

sult of several factors among the following: reduc-

tion of maintenance costs, of energy consumption, is-

sues related to purchasing software licenses and de-

preciation allowances now in charge of the provider,

just to mention some. The performance and secu-

rity/privacy characteristics are issues also prioritized

by users. With regard to security in a natural dis-

aster event,for example, the servers continue operat-

ing normally since the security policy of the service

providers located kilometers away from the headquar-

ters. Regarding the performance, it is possible to al-

locate resources in the cloud changing settings in the

provider and thereby allowing quick response to busi-

ness needs while increasing performance at times of

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

peak processing. This is an important factor to lead

to considerable stability in the services provided by

a company. Moreover, due to elasticity, ﬂexibility

and interoperability, previous demand resources can

return to their original conﬁguration, contributing to

possible reduction of costs. The infrastructure allo-

cated to a speciﬁc instance (e.g. memory, storage and

CPU) can vary according to different priorities de-

pending on the proﬁle of the type of service provided

to the ﬁnal user.

Table 3: Study evidence Consolidation.

Characteristics Studies

Studies

Quantity

Cost

S2, S9, S13,

S16, S32, S40,

S45, S50, S54,

S55, S57

Performance

S11, S44, S50,

S54, S55, S57

Security and Privacy

S25, S32, S46,

S49, S57

Infrastructure (Memory)

S13, S56, S57 03

Infrastructure (CPU)

S13, S56, S57 03

Infrastructure (Storage)

S56, S57 02

Flexibility S31, S32 02

Agility S57 01

Auditability S49 01

Accountability S57 01

Scalability S31 01

Elasticity S57 01

Effort S9 01

Quality Assurance

S57 01

Governability S49 01

Infrastructure (Network)

S56 01

Interoperability S49 01

Timeframe S32 01

Business Popularity

S32 01

IT Skilled Staff S32 01

Business Size S32 01

Usability S57 01

Figure 9: Characteristics Identiﬁed in the Studies.

Type of Study and Analysis Method. In Figure 10,

we can observe that studies use with more frequency

the strategy based on experience report. This is an

interesting ﬁnding that show possibilities to share ex-

perience among potentially interested parties in issues

related to cloud computing adoption and migration.

Figure 10: Types of Study.

Figure 11: Analysis Method.

On the other hand, in Figure 11, we can observe

a predominance of studies reporting difﬁculties, chal-

lenges and results obtained together with benchmark-

ing and difﬁculties analysis. This lead to 63% of the

occurrences in the studies, what brings possibilities

of lessons learnt in the community that are prone to

adopt the cloud computing paradigm.

6 THREATS TO VALIDITY

The following types of validity issues were consid-

ered when interpreting the results from this charac-

terization. Conclusion validity. There may be bias in

data extraction. However, this was addressed through

deﬁning a data extraction form to ensure consistent

A Characterization of Cloud Computing Adoption based on Literature Evidence

extraction of relevant data to answering the research

questions. The ﬁndings and implications are based

on the extracted data. Internal validity. One possi-

ble threat is the selection bias. We have addressed

this threat during the selection step as described in

Figure 1, i.e. the studies included in the character-

ization were identiﬁed through a thorough selection

process which comprises of multiple phases. Con-

struct validity. The studies identiﬁed from a previ-

ous SLR conducted by the authors were accumulated

from multiple literature databases covering relevant

journals and proceedings. One possible threat is bias

in the selection of publications. This was addressed

through specifying a research protocol that deﬁnes the

research questions and objectives of the study, inclu-

sion and exclusion criteria, search strings that we in-

tend to use, the search strategy and strategy for data

extraction.

External Validity. The set of 19 papers selected in

Phase 4 as described in Figure 1 is a potential ex-

ternal validity threat. In this case, there is a threat

that the results so far obtained could not be general-

ized. However, the studies were selected having as a

start point the ones published between 2005 and 2015

from a previous SLR conducted by the authors. For

this reason, the set of the 19 studies are considered

representative enough as a sampling for this charac-

terization.

7 CONCLUSIONS

The analysis of data and the consequent identiﬁca-

tion of strategies, approaches and tools reported in the

studies, could help in establishing knowledge on how

the companies should adopt and migrate to the cloud,

how the cost-beneﬁt relationship can be evaluated as

well as providers can be selected.

The selection of commercial cloud providers is

a challenging task and depends on several factors.

Among other reasons, cloud providers continually

upgrade their hardware and software infrastructures.

The result is that new commercial Cloud services,

technologies and strategies gradually enter the market

(Li et al., 2013). Studies have shown that successful

migration to the cloud are usually driven by a set of

criteria to select providers that best ﬁt the company

needs (Li et al., 2012b) (Li et al., 2010) (Garg et al.,

2013). According to the results of the characteriza-

tion presented in this paper, there is a tendency in the

studies for the public deployment model (1.1). An-

other important ﬁnding is the perception of cost re-

duction. This fact is associated with the absence of

the requirement to tie-up capital, to deal with techno-

logical obsolescence, hardware maintenance, as well

as purchasing software licenses and depreciation al-

lowances. In the cloud paradigm, these issues are now

in charge of the provider. We also identiﬁed a ten-

dency for the use of the IaaS model service. This can

be explained due to the relative less migration com-

plexity to the cloud that in this case is supported by

virtual machines. On the other hand, the adoption of

the PaaS model service has a potential drawback re-

ported in the studies the need to adapt the application,

including the need to rewrite parts of the code, re-

place libraries and APIs that can be not compatible

with the cloud provider environment. All these fac-

tors together contribute to a higher migration effort

and cost. However, the studies also highlighted the

need to evaluate the cost-beneﬁt relationship of both

possibilities: IaaS and PaaS. Another ﬁnding was

the emphasis of the studies in experience report for-

mat revealing opportunities of lessons learnt sharing.

The opportunities of cloud migration and provider se-

lection were conducted considering issues related to

cost, performance and security/privacy. To sum up,

it was possible to identify a tendency in the simulta-

neous use of public providers deployment model, the

IaaS service model and the Amazon platform. As a

future work, we plan to conduct a survey focusing on

companies that migrated to the cloud to conﬁrm the

results of this characterization study.

REFERENCES

Brereton, P., Kitchenham, B. A., Budgen, D., Turner, M.,

and Khalil, M. (2007). Lessons from applying the sys-

tematic literature review process within the software

engineering domain. Journal of systems and software,

80(4):571–583.

Buyya, R., Yeo, C. S., Venugopal, S., Broberg, J., and

Brandic, I. (2009). Cloud computing and emerging

it platforms: Vision, hype, and reality for delivering

computing as the 5th utility. Future Generation com-

puter systems, 25(6):599–616.

Chang, Y.-W. and Hsu, P.-Y. (2016). Investigating switch-

ing intention to cloud enterprise information systems:

an analysis at the organizational level. International

Journal of Information Management.

Garg, S. K., Versteeg, S., and Buyya, R. (2013). A frame-

work for ranking of cloud computing services. Future

Generation Computer Systems, 29(4):1012–1023.

Hashmi, S. I., Clerc, V., Razavian, M., Manteli, C., Tam-

burri, D. A., Lago, P., Di Nitto, E., and Richardson, I.

(2011). Using the cloud to facilitate global software

development challenges. In 2011 IEEE Sixth Inter-

national Conference on Global Software Engineering

Workshop, pages 70–77. IEEE.

Hurtaud, S. and de la Vaissire, L. (2011). How to ensure

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

control and security when moving to saas/cloud appli-

cations.

Kitchenham, B. and Charters, S. (2007). Guidelines for per-

forming systematic literature reviews in software en-

gineering. In Technical report, Ver. 2.3 EBSE Techni-

cal Report. EBSE.

Lee, S.-G., Chae, S. H., and Cho, K. M. (2013). Drivers

and inhibitors of saas adoption in korea. International

Journal of Information Management, 33(3):429–440.

Li, A., Yang, X., Kandula, S., and Zhang, M. (2010). Cloud-

cmp: comparing public cloud providers. In Proceed-

ings of the 10th ACM SIGCOMM conference on Inter-

net measurement, pages 1–14. ACM.

Li, Z., O’Brien, L., Cai, R., and Zhang, H. (2012a). Towards

a taxonomy of performance evaluation of commercial

cloud services. In Cloud Computing (CLOUD), 2012

IEEE 5th International Conference on, pages 344–

351. IEEE.

Li, Z., O’Brien, L., Zhang, H., and Cai, R. (2012b). On a

catalogue of metrics for evaluating commercial cloud

services. In Proceedings of the 2012 ACM/IEEE 13th

International Conference on Grid Computing, pages

164–173. IEEE Computer Society.

Li, Z., Zhang, H., OBrien, L., Cai, R., and Flint, S. (2013).

On evaluating commercial cloud services: A sys-

tematic review. Journal of Systems and Software,

86(9):2371–2393.

Sadighi, M. (2014). Accounting System on Cloud: A Case

Study. In 2014 11th International Conference on In-

formation Technology: New Generations, pages 629–

632. IEEE.

Stieninger, M., Nedbal, D., Wetzlinger, W., Wagner, G.,

and Erskine, M. A. (2014). Impacts on the Organiza-

tional Adoption of Cloud Computing: A Reconceptu-

alization of Inﬂuencing Factors. Procedia Technology,

16:85–93.

Weiss, A. (2007). Computing in the clouds. Computing, 16.

APPENDIX

Table 4: Studies included in the review.

ID Author, Title Venue Year

S2 A. Khajeh-Hosseini, D. Green-

wood, J. W. Smith and I. Som-

merville, The Cloud Adoption

Toolkit: Supporting Cloud Adop-

tion Decisions in the Enterprise.

Top Ten Cited Paper According to

Google Scholar

SPE 2012

S8 O. Sefraoui, M. Aissaoui and M.

Eleuldj, Cloud computing migra-

tion and IT resources rationaliza-

tion.

ICMCS 2014

S9 J. Alonso, L. Orue-Echevarria, M.

Escalante, J. Gorronogoitia and D.

Presenza, Cloud modernization

assessment framework: Analyz-

ing the impact of a potential mi-

gration to Cloud.

MESOCA 2013

Table 4: Studies included in the review (cont.).

ID Author, Title Venue Year

S11 L. Zhou, CloudFTP: A Case

Study of Migrating Traditional

Applications to the Cloud.

ISDEA 2013

S13 S. N. Srirama, V. Ivanistsev, P.

Jakovits, and C. Willmore, Direct

migration of scientiﬁc computing

experiments to the cloud.

HPCSim 2013

S16 Q. H. Vu and R. Asal, Legacy Ap-

plication Migration to the Cloud:

Practicability and Methodology.

SERVICES 2012

S25 A. Michalas, N. Paladi and C.

Gehrmann, Security aspects of e-

Health systems migration to the

cloud.

HealthCom 2014

S31 T. Boillat and C. Legner,

Why Do Companies Migrate

Towards Cloud Enterprise

Systems? A Post-Implementation

Perspective.

CBI 2014

S32 M. Sadighi, Accounting System on

Cloud: A Case Study.

ITNG 2014

S40 J. Garca-Galn, P. Trinidad, O. F.

Rana and A. Ruiz-CortsAutomated

conﬁguration support for infras-

tructure migration to the cloud.

FGCS 2015

S44 P. J. P. da Costa and A. M. R.

da Cruz. Migration to Windows

Azure Analysis and Comparison.

PROTCY 2012

S45 J. Knsemller and H. Karl A game-

theoretic approach to the ﬁnan-

cial beneﬁts of infrastructure-as-

a-service.

FGCS 2014

S46 H. Mouratidis, S. Islam, C. Kallo-

niatis and S. Gritzalis. A frame-

work to support selection of cloud

providers based on security and

privacy requirements.

JSS 2013

S49 C. Tang and J. Liu.Selecting a

trusted cloud service provider for

your SaaS program.

COSE 2015

S50 F. CRowe, J. Brinkley and N.

Tabrizi.Migrating Legacy Appli-

cations to the Cloud.

CLOUDCOM 2013

S54 R. Prodan and S. Ostermann. A

Survey and Taxonomy of Infras-

tructure as a Service and Web

Hosting Cloud Providers. Top

Ten Cited Paper According to

Google Scholar

IWGC 2009

S55 A. Li, X. Yang, S. Kandula and M.

Zhang. CloudCmp: Comparing

Public Cloud Providers. Top Ten

Cited Paper According to Google

Scholar

IMC 2010

S56 Z. Li, L. OBrien, R. Ranjan and

M. Zhang. Early Observations on

Performance of Google Compute

Engine for Scientiﬁc Computing.

CLOUDCOM 2013

S57 S. K. Garg, S. Versteeg and R.

Buyya. A framework for rank-

ing of cloud computing services.

Top Ten Cited Paper According to

Google Scholar

FGCS 2013

A Characterization of Cloud Computing Adoption based on Literature Evidence