BIG DATA MINING AND ANALYTICS

ISSN 2096-0654 02/10 pp127 –138

Vo l u m e 6 , N u m b e r 2, J u n e 2 0 2 3
DOI: 10.26599/BDMA.2022.9020016

Cloud-Based Software Development Lifecycle: A Simplified Algorithm

for Cloud Service Provider Evaluation with Metric Analysis
Santhosh S and Narayana Swamy Ramaiah

Abstract: At present, hundreds of cloud vendors in the global market provide various services based on a customer’s
requirements. All cloud vendors are not the same in terms of the number of services, infrastructure availability,
security strategies, cost per customer, and reputation in the market. Thus, software developers and organizations
face a dilemma when choosing a suitable cloud vendor for their developmental activities. Thus, there is a need to
evaluate various cloud service providers (CSPs) and platforms before choosing a suitable vendor. Already existing
solutions are either based on simulation tools as per the requirements or evaluated concerning the quality of service
attributes. However, they require more time to collect data, simulate and evaluate the vendor. The proposed work
compares various CSPs in terms of major metrics, such as establishment, services, infrastructure, tools, pricing
models, market share, etc., based on the comparison, parameter ranking, and weightage allocated. Furthermore, the
parameters are categorized depending on the priority level. The weighted average is calculated for each CSP, after
which the values are sorted in descending order. The experimental results show the unbiased selection of CSPs
based on the chosen parameters. The proposed parameter-ranking priority level weightage (PRPLW) algorithm
simplifies the selection of the best-suited cloud vendor in accordance with the requirements of software development.

Key words: cloud-based software development life cycle (SDLC); cloud evaluation; parameter-ranking priority level
weightage (PRPLW) algorithm; cloud service providers; software engineering

1 Introduction In this study, we analyzed the significant challenges

of software development in a cloud environment[1]
1.1 Cloud computing environment
and developed a cloud-based software development
There have been many improvements since the life cycle model as shown in Fig. 1[2] , thus helping
introduction of the cloud computing concept. Most users developers to utilize the model as a framework for cloud-
and organizations are interested in the cloud computing based developmental activities. The major challenges
model for their business activities, and software in selecting the cloud service providers (CSPs) are
developers and their organizations are not an exception in as follows: (1) the metrics to be considered for CSP
this case. Nowadays, users and organizations, do not face selection, (2) the various services offered by CSPs, and
a dilemma on whether to choose cloud or on-premise (3) the process of choosing the best, most suited CSP
resources. from the numerous cloud vendors existing in the global
 Santhosh S and Narayana Swamy Ramaiah are with Computer market.
Science & Engineering, Jain (Deemed-to-be University), 1.2 CSPs
Bangalore 562112, India. E-mail: [email protected];
[email protected]. There are numerous CSPs across the globe with different
* To whom correspondence should be addressed. services, pricing models, and infrastructures. The list of
Manuscript received: 2022-04-05; Revised: 2022-05-30: major CSPs across the globe is shown in Table 1.
Accepted: 2022-06-20 As per Gartner’s Magic Quadrant for Cloud

C The author(s) 2023. The articles published in this open access journal are distributed under the terms of the
Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/).
 128 Big Data Mining and Analytics, June 2023, 6(2): 127–138

Fig. 1 Design of cloud-based software development life cycle (SDLC).

Table 1 Major service providers in the global market of (2) Microsoft Azure,
cloud. (3) Google Cloud Platform (GCP).
Serial No. Cloud service provider Once the top three vendors are identified, the primary
1 Amazon web service (AWS)
goal is to choose the most appropriate one among them
2 ServerSpace
for a software requirement specification.
3 Microsoft Azure
4 Google Cloud Platform (GCP) 1.3 Motivation
5 IBM Cloud Services
6 Adobe Creative Cloud  Most of the existing solutions and methods talk about
7 Kamatera simulation tools for choosing service providers, and
8 VMware there is no clear comparison of services from top vendors
9 Rackspace to allow prospective users to choose the particular
10 Red Hat services they need while developing their software
11 Salesforce products.
12 Oracle Cloud
 Cost comparisons of the main CSPs and their various
13 SAP
pricing models are not available, furthermore these
14 Verizon Cloud
15 Navisite metrics cannot be found in one place.
16 Dropbox  Major existing solutions prefer to rank cloud vendors
based on quality of service (QoS) attributes such as
Infrastructure and Platform Services published in performance, throughput. Although the QoS attributes
2021[3] , as shown in Fig. 1, the top three CSPs are: play a vital role, the competing vendors may have
(1) Amazon Web Services (AWS), performance variation of ˙5% which should not cost
 Santhosh S et al.: Cloud-Based Software Development Lifecycle: A Simplified Algorithm for Cloud Service : : : 129

the precious time of developers. for systematically and succinctly representing QoS
1.4 Main contribution attributes[7] . Compared to other existing methods, this
framework fetches information from multiple sources
 This work clearly gives the comparison of various such as information from cloud vendors, performance
services offered, the costs involved, infrastructure monitoring data collected from multiple third-party tools,
requirements, various tools offered as a service, and more relevant reviews from users. The collected
and information about service availability and non- information from various sources helps in improving the
availability with respect to three cloud vendors for the selection process of cloud vendors. The model, however,
evaluation of the cloud by the developers, along with does not provide any information about the various
the advantages and disadvantages of individual cloud services offered by the vendors and the selection process
vendors. only depends on performance and reviews.
 A simplified algorithm named parameter-ranking Reference [8] used a neo-institutional lens and
priority level weightage (PRPLW) algorithm is proposed adopted several strategies (i.e., efficiency-motivated
to help developers choose the most suitable CSP from adoption, non-adoption, and legitimacy-motivated
the given choices. adoption) to understand the internal and external factors
 The proposed algorithm has been proven to work that impact adopting cloud computing. The work is
based on the parameter ranking and weightage allocated limited to a specific country and only considers factors
and the priority level weightage allocated after a that are limited to a particular region. Thus, more
thorough analysis of parameters. detailed research may be considered to deal with
individual factor analysis, and different countries also
2 Review of the Literature need to be examined.
The selection of the best and most suitable CSP Reference [9] proposed a method called “cloud service
from multiple vendors based on the specifications and research and selection system” (CSRSS), which selects
requirements is a challenging task for cloud users[4] . the service provider concerning QoS attributes specified
Upadhyay et al.[4] proposed a systematic QoS evaluation as per the requirements. Here the user must specify
and ranking framework for the cloud environment which the QoS attribute values depending on the requirements,
evaluated cloud vendors in terms of multiple qualitative although the proposed method lacks the ability to choose
and quantitative data and ranked them. The major the services required, and does not provide a cost
drawback is that collecting previous data for evaluation estimation.
purposes takes a longer time. Finally, Ref. [10] proposed a model whose design
Meanwhile the cloud services trust evaluation model is based on the client-CSP relationship. The three
(CSTEM) has been proposed to gain the trust of the major factors considered in choosing the CSP are client
users concerning security-related issues in the cloud satisfaction, relationalism, and dependence. The model
and to improve the reputation of an organization[5] . The was tested for one particular context, and the simulation
proposed model, however, has the following limitations. results were satisfactory. The model must be studied in
(1) The dynamic update factors must be considered terms of multiple contexts, and the results have to be
for cloud evaluation to improve the CSTEM model; further analyzed, especially the boundary conditions for
(2) The security services and performance metrics the proposed model.
have to be thoroughly experimented with, to improve the
3 Challenge
interaction success rate.
Wibowo et al.[6] proposed a performance evaluation Based on the review of literature, different cloud
method named the “fuzzy multicriteria group decision- computing adoption techniques face different challenges
making method” for cloud services. The interval-valued especially the data collection process, the simulation
intuitionistic fuzzy numbers are used to model the tool-based selection of cloud vendors, and the
evaluation process. Although this method proved to comparison of the current services offered, cost per
be an efficient model for ranking the service provider, it customer, tools required, and so on. When the QoS
is limited only to evaluating performance metrics. attributes are considered for evaluating the cloud vendor,
A previous study introduced a QoS-aware selection the performance may vary by ˙5%, which can be
model that developed a three-level representation scheme maximized by using high-capacity instances. The major
 130 Big Data Mining and Analytics, June 2023, 6(2): 127–138

challenges to be addressed for a user to choose the CSP offered in the categories of infrastructure such as virtual
are as follows: machines (EC2), storage options (Amazon S3), etc.
(1) The simplest algorithm to be followed while From 2009 onwards it introduced additional services,
choosing the CSP; such as Elastic Block Store, Amazon CloudFront, AWS
(2) The metrics to be considered for evaluating the Route 53, and AWS notification and messaging services.
CSP and their visibility to the evaluator; AWS currently holds a 32% market share across the
(3) Various features available from the vendors and globe and is considered the top-ranked CSP in the market
their pricing models; and share chart.
(4) A comparative study of different vendors based on Microsoft Azure initially started with the name
the metrics chosen. “Azure” in 2010. This CSP aims to provide various cloud-
based services from Microsoft to any other businesses. In
4 Comparative Study of the Top Three 2014, it was renamed “Microsoft Azure”. From the time
Cloud Vendors of its introduction, it has shown a very good progress
The identified top three cloud vendors were analyzed rate and competitive advantage. Currently, Azure has
based on the selected metrics, including cloud 19% of the global market share.
establishment, services offered, key tools offered as Google cloud platform started its cloud services in
part of the services, cloud platform, and pricing models. 2011. In just ten years, it has grown to become the
These basic metrics allow prospective users to assess the top three global CSP and has created an impact in the
vendors and choose the most suitable for them in terms market. Initially this service was launched to expand the
of software specification and development requirements. infrastructure to support YouTube and Google’s search
engine. GCP currently has 7% of the global market
4.1 Metrics for comparison share.
4.1.1 Cloud establishment and market share 4.1.2 Cloud services
Amazon web services is a pioneer in the area of cloud The cloud providers offer a variety of services and are
computing and is an experienced service provider in the broadly categorized as computing, storage, database,
market. AWS has major clients using its services with and networking services[11] . The comparison of services
trust and satisfaction. offered by the top three cloud vendors namely AWS[12] ,
Initially, AWS launched cloud services for its e- Azure[13] , and GCP[14] , for these categories are shown in
commerce business purpose. Later in the year 2006 Tables 2–5.
it started providing services publicly with major services By considering the overall services, AWS, Azure, and

Table 2 Computing services.

Cloud service Amazon web services Microsoft Azure Google cloud platform
Infrastructure as a service Amazon EC2 (Elastic Compute Cloud) Virtual machines Google Compute Engine
Platform as a service AWS Elastic Beanstalk App service and cloud services Google App Engine
Container options Amazon EC2 Container Service Azure Kubernetes Service (AKS) Google Kubernetes Engine
Serverless functionality AWS Lambda Azure Functions Google Cloud Functions

Table 3 Database services.

Cloud service Amazon web services Microsoft Azure Google cloud platform
Relational database management system Amazon Relational Database Service SQL database Google Cloud SQL
Google Cloud Datastore
NoSQL: key–value Amazon DynamoDB Table storage
Google Cloud Bigtable
NoSQL: indexed Amazon SimpleDB Azure Cosmos DB Google Cloud Datastore

Table 4 Storage services.

Cloud service Amazon web services Microsoft Azure Google cloud platform
Object storage Amazon Simple Storage Service Blob Storage Google Cloud Storage
Virtual server disks Amazon Elastic Block Store Managed disks Google Compute Engine Persistent Disks
Cold storage Amazon Glacier Azure Archive Blob Storage Google Cloud Storage Nearline
File storage Amazon Elastic File System Azure File Storage ZFS/Avere
 Santhosh S et al.: Cloud-Based Software Development Lifecycle: A Simplified Algorithm for Cloud Service : : : 131

Table 5 Networking services.

Cloud service Amazon web services Microsoft Azure Google cloud platform
Virtual network Amazon Virtual Private Cloud (VPC) Virtual Networks (VNets) Virtual private cloud
Elastic Load balancer Elastic Load Balancer Load Balancer Google Cloud Load Balancing
Peering Direct Connect ExpressRoute Google Cloud Interconnect
Domain name system Amazon Route 53 Azure DNS Google Cloud DNS

the GCP provide approximately 200+, 100+, and up to 4.2 CSPs: Pros and Cons
60+ cloud services, respectively. Table 9 provides insights into the advantages and
4.1.3 Key cloud tools disadvantages of the top three major cloud vendors based
All three CSPs face huge competition in providing on the analysis of metrics considered in the current work.
cloud-based tools to end-users. For the current industry
5 Comparison of Availability and Non-
needs and customer requirements, several tools are
offered as services[15] by the three vendors, as shown in Availability of Services from Vendors
Table 6. It is very important for software developers to know
4.1.4 CSP pricing models which services are required for their development
The three CSPs use various pricing models to satisfy activity and at which particular period of the
user needs. A comparison of their pricing models is development plan they will be used. Therefore, the
presented in Table 7. developer must check for the availability of such services
The comparison shown in Table 8 provides insights required for development from the chosen vendor.
into pricing models for the virtual machine types[16] Table 10 provides insights into the available (marked
p
of AWS, Azure, and GCP. as / and non-available (marked as / services from
Table 6 Comparison of tools from various cloud vendors.
Cloud tools Artificial intelligence and machine learning Internet of things Serverless functionality
(1) IoT Device Management
(1) Tensorflow on AWS, Apache MXNet on (2) Greengrass
AWS (3) IoT Button
Amazon Web (2) Sagemaker, Deep learning Amazon Machine (4) IoT Device Defender (1) Lambda
Services Images, DeepLens (5) IoT Analytics (2) Serverless Application Repository
(3) Transcribe, Translate, Machine Learning (6) FreeRTOS
(4) Rekognition, Polly, Lex, Comprehend (7) IoT Core
(8) IoT 1-click
(1) Time Series Insights
(1) Cognitive Services
(2) IoT Edge
Microsoft Azure (2) Machine Learning Functions
(3) Stream Analytics
(3) Azure Bot Service
(4) IoT Hub
(1) Cloud Job Discovery (Private Beta),
Cloud Video Intelligence, Cloud Translation
API, Cloud Speech API
Google Cloud Cloud IoT Core (Beta) Cloud Functions (Beta)
(2) Dialogflow Enterprise Edition, Cloud
Platform
Natural Language
(3) Cloud Machine Learning Engine

Table 7 Pricing models of the top three CSPs.

Cloud service provider Pricing Model
AWS Per hour (rounded up) On-demand, spot, reserved
Per minute (rounded up)
Azure Per minute (rounded up) On-demand (short-term)
Prepaid type or monthly billing Prepaid type or monthly billing
GCP Per minute (rounded up, minimum 10 mins) On-demand (sustained use)
Per second (rounded up)
 132 Big Data Mining and Analytics, June 2023, 6(2): 127–138

Table 8 Pricing models of the top three CSPs at the virtual machine level.
Virtual machine type AWS Azure GCP
Smallest instance vCPUs: 2 vCPUs: 2 vCPUs: 2
RAM: 8 GB RAM: 8 GB RAM: 8 GB
Cost: USD 69 per month Cost: USD 70 per month Cost: USD 52 per month
Largest instance vCPUs: 128 vCPUs: 128 vCPUs: 160
RAM: 3.84 TB RAM: 3.89 TB RAM: 3.75 TB
Cost: USD 3.97 per hour Cost: USD 6.79 per hour Cost: USD 5.32 per hour

Table 9 Pros and cons of top three cloud vendors.

Cloud service providers Pros Cons
Amazon web services (1) Oldest and strongest hold in the cloud market (1) Little difficulty for customers to use
(2) Extensive and mature service offerings (2) Managing the cost per customer
(3) Support for large-scale organizations as the leader (3) Multiple services and varieties of options are
in infrastructure available, so searching for suitable services can be
(4) Provides necessary training to customers challenging
(5) Higher reachability across the globe
Microsoft Azure (1) Top two ranked CSP (1) The documentation process has issues
(2) Easy to integrate the software and tools with (2) The tools for managing the services are
Microsoft products insufficient
(3) Broad feature set
(4) Open-source support
Google Cloud Platform (1) Cost-efficient infrastructure service (1) Late entry to the cloud market compared to
(2) Option for portability other top vendors
(3) Open source support (2) Less number of services offered
(4) Flexibility in agreements and discount offers. (3) Infrastructure used is the same as the Google
(5) Expertise in development and operations search engine and YouTube facility.

the three major cloud vendors. products with on-premise systems and open-source
software, Azure is the better option.
6 PRPLW-Based CSP Selection Algorithm  Pricing models and cost efficiency. Based on user-
Based on the analysis with respect to metrics or friendly pricing models and the deep discounts offered
parameters considered, such as establishment, services, to the user, the GCP is a better choice.
infrastructure, tools, pricing models, and current market  Reputation. Based on the market share, brand name,
share, the three CSPs are ranked against each metric and utilization, AWS has the upper hand.
or parameter and assigned a weightage, as shown in  Storage service options.
Table 11. The parameter-ranking and weightage assigned ı AWS provides three storage types (object, block,
are represented in graphic format in Fig. 2. and file storage).
 Establishment. AWS is ahead of any other CSPs as ı Azure provides only two types of storage (object
it is the pioneer in the industry. and file storage).
 Infrastructure. For infrastructure, AWS clearly ı GCP provides only two types of storage (object
stands tall as it has more regions and availability zones. and block storage).
 Market share. Having 32% of the global share in  Archival and backup. AWS and Azure provide
the cloud market, AWS has the upper hand. archival storage with lifecycle management, whereas
 Growth rate. Comparatively, GCP is growing faster GCP does not have this option.
by almost 100%.  Offline data transfer. AWS leads with mobile
 Utilization. Most of the major customers of the devices for offline transfer of data.
cloud are using the top three CSPs almost equally. As far as the remaining parameters (e.g., availability,
 Services. disaster recovery, cloud tools, accountability, service
ı Considering the overall services offered, AWS response time, etc.) are concerned, all the three CSPs
offers 200+ services compared to Azure or GCP. are equally competent.
ı For the integration of services, such as Microsoft The parameter selection for individual projects differs
 Santhosh S et al.: Cloud-Based Software Development Lifecycle: A Simplified Algorithm for Cloud Service : : : 133

Table 10 Comparison of the availability and non-availability of services offered by the top three CSPs.
Cloud service Amazon web service Microsoft Azure Google cloud platform
p p p
Regions and availability zones
p p p
Various pricing models
p p p
Virtual machines for computing purpose
p p p
Containers services
p p p
Serverless functionality
p p p
Mobile app hosting service on the cloud
p p
Batch processing service 
p p p
Object storage
p p
Block storage 
p p
File storage 
p p
Hybrid storage 
p p
Offline data transfer 
p p p
Relational database service
p p p
NoSQL database
p p
In-memory database 
p p p
Archival data and backup service
p p
Disaster recovery management 
p p p
Machine learning tools and services
p p p
Cognitive services
p p p
IoT tools and resources service
p p p
Networking services
p p p
Content delivery network services
p p p
Big data analytics services
p p p
Identity access management and
authentication service
p p p
Cloud security services
p p
Application lifecycle management 
p p p
Cloud monitoring and management
p p p
AR & VR service
p p
VPC on public cloud 
p p p
Customer training service
p p p
Customer support service
p p p
Third-party software as a service

from one another. Each parameter contributes to the The selected parameters from the user are categorized
project and has its own priorities. The weightage of less as priority level 1 (PL1), priority level 2 (PL2), and
prioritized parameters should not influence the selection priority level 3 (PL3) based on Table 12 and the values
of CSPs. Thus the parameters are categorized into three for n1 ; n2 ; and n3 are determined (Number of parameters
priority levels (highest to lowest) and assigned constant in each category). The weightage will be assigned to
weightage based on the priority, as shown in Table 12. each level. Then the weighted average for each CSP is
In this study, the PRPLW algorithm is developed for calculated separately. The obtained weighted average
the selection of a suitable CSP based on the requirements values are sorted in descending order and then ranked.
of the software to be developed by considering the The top-ranked CSP shall be suggested as the suitable
parameters, rankings, and weightages in Tables 11 and CSP based on the chosen parameters.
12. The process diagram for the PRPLW-based CSP 6.1 Sample experimentations and results
selection algorithm (Algorithm 1) is shown in Fig. 3.
Before choosing the CSP, the software development The parameters are selected by the developer based
team must have the software requirement specifications on the requirements specifications. These parameters
and the important parameters concerning cloud-based are categorized in accordance with priority levels and
development activities as inputs to the algorithm. weightage allocated, as shown in Table 12. Based on the
 134 Big Data Mining and Analytics, June 2023, 6(2): 127–138

Table 11 Parameter-based ranking and weightage of the CSP selection.

Ranking Weightage
Parameter
AWS Azure GCP AWS Azure GCP
Establishment 1 2 3 1 0.75 0.5
Infrastructure 1 2 3 1 0.75 0.5
Number of services 1 2 3 1 0.75 0.5
Utilization 1 2 3 1 0.75 0.5
Pricing models and cost efficiency 2 3 1 0.75 0.5 1
Storage service options 1 2 3 1 0.75 0.5
Service response time 1 1 1 0.75 0.75 0.75
Accountability 1 1 1 0.75 0.75 0.75
Availability 1 1 1 0.75 0.75 0.75
Cloud tools 1 1 1 0.75 0.75 0.75
Archival and backup 1 2 3 1 0.75 0.5
Market share 1 2 3 1 0.75 0.5
Growth rate 3 2 1 0.5 0.75 1
Offline data transfer 1 2 3 1 0.75 0.5
Disaster recovery 1 1 – 0.75 0.75 0
SLA flexibility 1 1 1 0.75 0.75 0.75
Compatibility with Microsoft products 2 1 3 0.75 1 0.5
Reputation 1 2 3 1 0.75 0.5
Note: Weightage is assigned for each parameter of CSP based on the ranking as follows: 1 for rank 1, 0.75 for rank 2, and 0.5 for rank 3.

Fig. 2 Parameter-ranking and weightage.

parameter weightage and priority level weightage, the reputation, establishment, market share, and growth rate.
weighted average for each CSP is calculated separately  Step 2. Input parameters: cost efficiency,
and then sorted in descending order. The highest infrastructure, availability, SLA flexibility, utilization,
weighted average CSP is the suggested CSP for the growth rate.
chosen parameters.  Step 3. Categorization. PL1 category: cost efficiency,
6.2 Experiment 1 infrastructure, availability, SLA flexibility (n1 D 4);
 Step 1. Display parameters: pricing models and cost PL2 category: utilization (n2 D 1); PL3 category:
efficiency, infrastructure, wide range of services, service growth rate (n3 D 1).
response time, accountability, availability, archival and  Step 4. Assigning weightage to the following
backup, SLA flexibility, storage service options, offline categories: PL1 D 0:5, PL2 D 0:3, PL3 D 0:2:
data transfer, disaster recovery, cloud tools, utilization,  Step 5. Calculation of the weighted average using
 Santhosh S et al.: Cloud-Based Software Development Lifecycle: A Simplified Algorithm for Cloud Service : : : 135

Table 12 Priority level categories and weightage.

Parameter Priority level Weightage
Pricing models and cost efficiency
Infrastructure
Wide range of services
Service response time
PL1 (High) 0.5
Accountability
Availability
Archival and Backup
SLA flexibility
Storage service options
Offline data transfer
Disaster recovery
PL2 (Medium) 0.3
Compatibility with Microsoft products
Cloud tools
Utilization
Reputation
Establishment
PL3 (Low) 0.2
Market share
Growth rate

the following equations. The PRPLW algorithm sorted the weighted average
     
3:25 1 0:5 values and ranking, as shown in Table 14.
PAVG, AWS D 0:5C 0:3C 0:2 D
4 1 1 In this experiment, he top-ranked Azure Cloud
0.806 25 is suggested as the suitable CSP based on the

2:75
 
0:75
 
0:75
 chosen parameters for development. The graphical
PAVG, AZ D 0:5C 0:3C 0:2 D representation of the results is shown in Fig. 5.
4 1 1
0:718 75 The parameters selected are compatibility with
 
3
 
0:5
 
1 Microsoft products, disaster recovery, service response
PAVG, GCP D 0:5C 0:3C 0:2 D 0:725 time, cloud tools, availability, SLA flexibility, utilization,
4 1 1
 Step 6. Sorting the weighted average values and and growth rate. The compatibility for Microsoft
ranking, as shown in Table 13. products is provided by Azure Cloud without additional
 Step 7. Output: the top-ranked AWS is suggested configurations, while some of the parameters have the
as the suitable CSP based on the chosen parameters for same (or less) priority. Thus, Azure is the most suitable
development. The graphical representation of the results CSP for this user.
is shown in Fig. 4. 6.4 Experiment 3
As shown in the example, the chosen parameters by In this experiment, the input parameters are cost
the developer are infrastructure, availability, and cost efficiency, storage service options, cloud tools,
efficiency. The results show that AWS provides better accountability, SLA flexibility, and growth rate.
infrastructure compared to any other service provider, The PRPLW algorithm sorted the weighted average
with an availability of 99.99%. As per the weighted values and ranking, as shown in Table 15.
average calculation, the AWS cloud is thus suggested for In this experiment, the top-ranked GCP is suggested
the developer. as the suitable CSP for this user based on the
6.3 Experiment 2 chosen parameters for development. The graphical
In this experiment, the input parameters are compatibility representation of the results is shown in Fig. 6.
with Microsoft products, disaster recovery, service When the cost is the major factor compared to any
response time, cloud tools, availability, SLA flexibility, parameters, the Google Cloud Platform is the best-
utilization, and growth rate. The calculation of this suited cloud provider. The developer has chosen the cost
and the following experiments is similar to that of parameter along with other parameters which are having
Experiment 1. common priority. So the Google Cloud Platform is the
 136 Big Data Mining and Analytics, June 2023, 6(2): 127–138

growth rate.
Algorithm 1 PRPLW-based CSP selection algorithm
Input: Pi , the important parameters pertaining to software The PRPLW algorithm sorted the weighted average
development. values and ranking, as shown in Table 16.
Output: Po , the suitable CSP as per the requirements and In this experiment the top-ranked GCP is suggested
parameters. as the suitable CSP in this case based on the
Function: Selection of CSP chosen parameters for development. The graphical
1: Display the parameters for consideration as per the software
representation of the results is shown in Fig. 7.
requirements specification (SRS).
2: Input the various parameters chosen for software
The parameters chosen are cost efficiency, reputation,
development. establishment, market share, growth rate. All the
3: Categorize the parameters as per the following priority levels: parameters except cost suggest the AWS cloud based on
n1 as number of parameters in the PL1 category, n2 as number the parameter ranking, which is a biased result. When it
of parameters in the PL2 category, and n3 as number of comes to cost, the GCP is the most cost-efficient cloud
parameters in the PL3 category.
provider. The priority level and weightage allocation
4: Assign weightage for each priority (PL1 D 0:5; PL2 D
0:3; PL3 D 0:2). thus come into the picture to obtain an unbiased result.
5: Calculate the weighted average for the individual CSPs using Nevertheless, the weighted average calculation suggests
the following equation. GCP as the most suitable CSP.
n1 n2
X X The experimentation results show the selection of
PAVG, AWS D .Pi; AWS =n1 /  PL1 C .Pi; AWS =n2 /
i D1 iD1
CSPs based on the parameters chosen for development
n3
X using the PRPLW based CSP selection algorithm. The
PL2 C .Pi; AWS =n3 /  PL3 (1) experiment results are available for each CSP, and one
i D1
n1
X n2
X result is shown for an unbiased CSP selection.
PAVG, AZ D .Pi; AZ =n1 /  PL1 C .Pi; AZ =n2 /
iD1 iD1 7 Conclusion and Future Work
n3
X
PL2 C .Pi; AZ =n3 /  PL3 (2) Considering the above analysis, an important and
n1
X
i D1
n2
X
challenging task in the process of selecting the CSP
PAVG, GCP D .Pi; GCP =n1 /  PL1 C .Pi; GCP =n2 / for any software development activities on the cloud has
iD1 iD1 been simplified using the proposed PRPLW based CSP
n3
selection algorithm. By analyzing the parameters with
X
PL2 C .Pi; GCP =n3 /  PL3 (3)
i D1 the comparative study for all three CSPs, along with their
6: Sort the weighted average values in ascending order. ranking and allocated weightage a development team can
7: Output the rank as per the sorted order and suggest the top choose the parameters that have an impact on software
rank as the suitable CSP based on the chosen parameters.
development activities from the set of parameters listed.
The proposed algorithm shows unbiased results, as the
most suitable cloud provider.
introduction of priority level weightage removes any
6.5 Experiment 4 such bias. The experimentation results reveal that the
In this experiment, the input parameters are cost process of CSP selection is simplified and takes less time
efficiency, reputation, establishment, market share, and to decide on selection. Thus, the developer can select

Fig. 3 Process diagram for the PRPLW-based CSP selection algorithm.

 Santhosh S et al.: Cloud-Based Software Development Lifecycle: A Simplified Algorithm for Cloud Service : : : 137

Table 13 Weighted average values and ranking in Table 16 Weighted average values and ranking in
Experiment 1. Experiment 4.
Ranking Sorted value Cloud service provider Ranking Sorted value Cloud service provider
1 0.806 25 Amazon Web Services 1 0.625 Google Cloud Platform
2 0.725 Google Cloud Platform 2 0.525 Azure Cloud
3 0.718 75 Azure Cloud 3 0.425 Amazon Web Services

Fig. 4 Experiment 1 results. Fig. 7 Experiment 4 results.

Table 14 Weighted average values and ranking in addressing all CSPs, which must be refined with the use
Experiment 2. of additional metrics.
Ranking Sorted value Cloud service provider
1 0.852 08 Azure Cloud References
2 0.843 75 Amazon Web Services
[1] S. Santhosh and N. Ramaiah, The impact of software
3 0.747 91 Google Cloud Platform
engineering methods for cloud computing models – A
survey, https://dx.doi.org/10.2139/ssrn.3372019, 2019.
[2] S. Santhosh and N. S. Ramaiah, Understanding the
significant challenges of software engineering in cloud
environments, in Computational Intelligence Techniques
and Their Applications to Software Engineering Problems,
A. Bansal, A. Jain, S. Jain, V. Jain, and A. Choudhary, Eds.
Boca Raton, FL, USA: CRC Press, 2020.
[3] Raj Bala, Bob Gill, Dennis Smith, Kevin Ji, David Wright,
https://www.gartner.com/doc/reprints?id=1271OE4VR&ct=
Fig. 5 Experiment 2 results. 210802&st=sb& ga=2.218192091.335998994.1642577266–
132792268.1642577266, 2021.
Table 15 Weighted average values and ranking in [4] N. Upadhyay, Managing cloud service evaluation and
Experiment 3. selection, Procedia Comput. Sci., vol. 122, pp. 1061–1068,
Ranking Sorted value Cloud service provider 2017.
1 0.8875 Google Cloud Platform [5] Y. Wang, J. Wen, X. Wang, B. Tao, and W. Zhou, A cloud
2 0.820 83 Amazon Web Services service trust evaluation model based on combining weights
3 0.791 66 Azure Cloud and gray correlation analysis, Secur. Commun. Networks,
vol. 2019, p. 2437062, 2019.
[6] S. Wibowo, H. Deng, and W. Xu, Evaluation of cloud
services: A fuzzy multi-criteria group decision making
method, Algorithms, vol. 9, no. 4, p. 84, 2016.
[7] M. Eisa, M. Younas, K. Basu, and I. Awan, Modelling
and simulation of QoS-aware service selection in cloud
computing. Simul. Modell. Pract. Theory, vol. 103, p.
102108, 2020.
[8] R. El-Gazzar, E. Hustad, and D. H. Olsen, An institutional
lens on cloud computing adoption – A study of institutional
Fig. 6 Experiment 3 results. factors and adoption strategies, in Proc. 25th European
the most suitable CSP based on the software requirement Conf. Information Systems (ECIS), Guimarães, Portugal,
specification. However, evaluation metrics considered in 2017, pp. 2477–2492.
[9] M. Abourezq and A. Idrissi, Integration of QoS aspects in
this work concerning software requirement specifications the cloud service research and selection system, Int. J . Adv.
in relation to the top three CSPs identified. Thus, Comput. Sci. Appl., vol. 6, no. 6, pp. 111–122, 2015.
future works must propose general evaluation models [10] D. Schlagwein and A. Thorogood, Married for life? A cloud
 138 Big Data Mining and Analytics, June 2023, 6(2): 127–138

computing client–provider relationship continuance model, [14] Google Cloud, https://cloud.google.com/products/, 2022.
in Proc. 22nd European Conf. Information Systems, Tel [15] A. Bartwal, AWS vs Azure vs GCP & AWS vs Azure vs
Aviv, Israel, 2014. GCP Comparison Overview, https://k21academy.com/
[11] Intellipaat, AWS vs Azure vs Google Cloud: Choosing the amazon-web-services/aws-solutions-architect/aws-vs-azure-
Right Cloud Platform, https://intellipaat.com/blog/aws-vs- vs-gcp/, 2021.
azure-vs-google-cloud/, 2022. [16] C. Harvey, AWS vs. Azure vs. Google Cloud: 2022 cloud
[12] AWS Cloud Products, https://aws.amazon.com/products/, platform comparison, https://www.datamation.com/cloud/
2022. aws-vs-azure-vs-google-cloud/, 2021.
[13] Microsoft Azure, https://azure.microsoft.com/, 2022.

Narayana Swamy Ramaiah received worked as an expatriate faculty at the Ethiopian University and
the PhD degree from PRIST University, executed funded projects under MOHE, Ethiopia. Likewise,
Tamilnadu, India in 2016, the MTech he has participated in international conferences held in Oman,
degree from Visvesvaraya Technological Dubai, Singapore, and Bangkok. He has four international
University, Karnataka, in 2004, and the awards to his credit and has guided many undergraduate and
BE degree from Bangalore University, postgraduate students during their project completion throughout
Karnataka, in 2002. He has over 16 years his career.
of experience in teaching, research and
industry. He has published over 45 papers in peer-reviewed
journals. He is also a reviewer for peer-reviewed journals Santhosh S is pursuing the PhD at
published by Springer, IEEE, INSTICC, and Elsevier, and Jain Deemed-to-be University, Bangalore,
serves on the editorial board of Springer, Elsevier, SCIREA, India. He received the MTech (CSE)
IJETA, Bioinfo Publications journals, and the International degree from Visvesvaraya Technological
Journal of Networking. He has written book chapters on University, Karnataka, India, in 2015, and
several topics, namely, programming language, cryptography and the BE (CSE) degree from Anna University,
network security, Java, high-speed VLSI design, management, Tamilnadu, India, in 2009. Overall, he
and entrepreneurship. His areas of research include IoT has 12 years of teaching and industry
(agriculture), blockchain, AI, machine learning, and cloud experience. He has published three papers in peer-reviewed
computing. He is also a member of IEEE (Sensor Council and journals and presented four papers at international conferences
IoT), LMISTE, INSTICC Membership, SSRN–Elsevier. He and four papers at national conferences. His research areas include
is also an SPIE member and life member for ISTE. He has cloud computing and software engineering.