2021 International Conference on Technological Advancements and Innovations (ICTAI)

Cloud Computing and Comparison based on Service

and Performance between Amazon AWS, Microsoft
Azure, and Google Cloud
2021 International Conference on Technological Advancements and Innovations (ICTAI) | 978-1-6654-2087-7/21/$31.00 ©2021 IEEE | DOI: 10.1109/ICTAI53825.2021.9673425

Prakarsh Kaushik Ashwin Murali Rao Devang Pratap Singh

Student of Computer Science and Student of Computer Science and Student of Computer Science and
Engineering Engineering Engineering
Amity School of Engineering and Amity School of Engineering and Amity School of Engineering and
Technology Technology Technology
Greater Noida, India Greater Noida, India Greater Noida, India
[email protected] [email protected] [email protected]

Swati Vashisht Shubhi Gupta

Department of Computer Science Department of Computer Science Amity
Amity School of Engineering and School of Engineering and
Technology Technology
Greater Noida, India Greater Noida, India
[email protected] [email protected]

Abstract— Cloud computing is the on-request supply of capacity to scale elastically is one of the advantages of cloud
computing resources through the Web with pay-as-you-use computing services. In cloud parlance, this means delivering
billing. Instead of purchasing, operating, and maintaining the appropriate quantity of computing resources as most
physical computers, hardware, and servers, cloud solutions cloud computing solutions are self-service and on-demand,
providers such as Microsoft Azure of Microsoft, Amazon Web
even massive volume of computing power may be delivered
Services (AWS) of Amazon, and Google Cloud Platform (GCP)
by Google offer cloud solutions such as processing power, in moments, generally with only a few keystrokes and taps
memory, and databases on an as-needed basis. This research on mouse, allowing enterprises a lot of flexibility and
paper discusses the architecture and types of cloud computing relieving capacity planning strain. In addition to these, most
services, as well as comparison of the performance and service cloud providers provide such a comprehensive set of terms,
among three main Cloud Computing platforms: Microsoft technologies, and restrictions that increases the user's overall
Azure, Amazon AWS, and Google Cloud Platform. All three security posture, assisting in the protection of data, apps, and
systems have been evaluated in identical virtual environments, architecture from cyber threats.
specifically micro instance of Ubuntu 16.04. The benchmark
application Phoronix Test Suite 10.4 is used to assess
Cloud computing is the ultimate successor of virtualization, a
performance, and the results for the Apache, Dbench, and RAM
speed benchmarks are evaluated in this paper. technology based on service-based infrastructure that
provides the user with the abstraction and encapsulation of
Keywords— Cloud, Cloud Computing, Amazon AWS, Microsoft software and hardware parts in order to use them productively
Azure, Google Cloud, Virtual Environment, Virtualization for specific needs. Virtualization provides for more suitable
use of the system by allowing it to be used by multiple
I. INTRODUCTION operating systems at the same time. When used in
conjunction with cloud computing, it provides a more safe
Cloud computing in simple terms is providing computer and stable environment, owing mostly to VM isolation. A
services such as servers, storage, databases, networking, User of cloud solutions do not need to be knowledgeable of
software, analytics, and intelligence over the Web in order to system details, since they can access documents and
provide innovation at pace, more flexible resources, and resources through a basic web browser account to the
economies of scale. A user pays for the cloud solution that operator's cloud interface. The most significant advantage of
they use, allowing them to reduce operational expenses, run this technique is the ease of having a group of data centers in
infrastructure more productively, and scale as their enterprise one location, while the cost of using these resources is several
needs to expand. times less than the cost of operating one's personal computing
infrastructure.
Cloud computing represents a significant departure from the
typical way that enterprises think about computing resources. A data center is suitable for businesses which require a
It removes the investment cost of purchasing hardware and tailored, specialized solution that offers them complete
software, as well as establishing and operating on-site control over their resources and hardware. A data center, on
datacenters—server racks, the 24 hours electricity for power the other hand, has finite capacity; if a corporation constructs
and cooling, and IT specialists to manage the system. The a data center, it will be unable to adjust the storage capacity

268
978-1-6654-2087-7/21/$31.00 ©2021 IEEE

Authorized licensed use limited to: Beirut Arab University. Downloaded on February 22,2024 at 08:57:05 UTC from IEEE Xplore. Restrictions apply.
2021 International Conference on Technological Advancements and Innovations (ICTAI)

and load it can withstand without spending on and The Computing infrastructure, which relies on a variety of
maintaining additional tools. A cloud system, on the other applied technologies to enforce the abstraction of physical
hand, maybe scaled to meet the needs of a company's resources through virtualization and their subsequent
operations. Based on the seller's products and service options, dissemination to diverse users, is the foundation of cloud
it has theoretically infinite capacity. One downside of the computing. Two elements serve as the foundation of the
cloud solution is that the user does not have as much power Cloud computing architectural paradigm.
as they would in a data center because the system is managed
by a third party. A. Frontend – The client-side (client infrastructure) of cloud
computing is referred to as the frontend of the cloud
architecture. That is, it contains all of the UI and
applications that the client uses to access cloud resources.
For example, to access the cloud platform, you may need
the help of a web browser.

B. Backend - The cloud itself, as used by the service

provider, is referred to as the backend. It holds the
resources, maintains them, and incorporates security
procedures. It also contains massive storage, virtual
applications, virtual computers, traffic management
techniques, deployment models, etc.

1. Application – It is the platform that provides the cloud

service to the user as per their requirements.
Figure 1
2. Storage - Provides scalable storage service and
II. SIMILAR RESEARCH IN THE FIELD OF CLOUD management of data in cloud as per user requirements.

Over the last decade, there's been a great deal of study and 3. Service - There are mainly three major types of cloud
work in the developing field of cloud computing. One of the services IaaS, PaaS, and SaaS.
most pertinent problems is the provision of methodology and
quantitative evaluation of service quality and performance a. IaaS – IaaS - Infrastructure as a service (IaaS) is a
parameters among various cloud computing service solution delivery model that entails the leasing of a
providers. Some researchers raised the issue of potential company’s hardware to support the business of other
interference among Cloud users when pooling resources, and customers, which includes storage space, hardware,
as a result, they presented various figures of measurement to servers, and network components. Users of IaaS
assess the performance isolation of cloud-based systems. often spend money for services on a per-use basis.
Apart from that they define novel methods for attaining Virtual computers, storage space, firewalls, and load
performance isolation in cloud computing. Data centers balancers are all available from operators. These
consume a significant amount of energy, which is predicted things can be leased on demand, and the user has no
to increase significantly in line with current technological control over or maintenance of the cloud
advancements. As a result, a number of studies have been architecture, only managing the on-site OS, software
carried to address the developing issue of delivering an upgrades, and apps used by user.
energy-efficient cloud solution. To address the dispute of
cloud resource scheduling optimization, few researchers b. PaaS - PaaS - Platform as a service (PaaS) is a
suggested a specific cost-effective solution for service solution delivery paradigm that allows users to
request scheduling in cloud computing, with the purpose of lease virtualized machines along with related
processing dynamic user service requests more effectively in facilities for the use of current software as well
terms of price and without violating any Service Level as for the production, quality test, and use of new
Agreement (SLA). Some studies additionally look into softwares. Unlike traditional systems, however, PaaS
design enhancement, policy related to data security, and provides a environment to produce scalable
method enforcement. This paper compares the fulfillment and applications at cheap price.
service quality of three cloud platforms: Amazon AWS,
Microsoft Azure, and Google Cloud. c. SaaS - Software as a service (SaaS) is a software
model used for distribution of software which is
III. THE ARCHITECTURE AND CATEGORIES OF based on the availability and global accessibility of
CLOUD SERVICES programs hosted by service providers to users over a
specialized network or web. This technique enables

269

Authorized licensed use limited to: Beirut Arab University. Downloaded on February 22,2024 at 08:57:05 UTC from IEEE Xplore. Restrictions apply.
2021 International Conference on Technological Advancements and Innovations (ICTAI)

for the most efficient working of software Microsoft's cloud services and resources. These services and
applications, smooth administration, lower pricing, resources include storing and altering data based on user
and the elimination of the need for obtaining needs. All a user needs to gain access to these resources and
software license, physical installation, maintenance, services is an established internet connection and the ability
and buying the hardware resources required for the to log on to the Azure portal. It was launched in 2010, and
programs to function properly. The flexibility now is among the top cloud service providers and competitor
obtained by splitting the workload across a group of to AWS and GCP. Azure offers over 100+ cloud computing
servers is one of the most significant advantages of services.
SaaS over traditional applications.
For their cloud computing requirements, 80 percent of
4. Cloud Runtime – It refers to the provision of executing Fortune 500 organizations use Azure. eBay, Samsung, HP,
and running environments to virtual machines. and other well-known corporations are on this list. It is
compatible with a variety of programming languages,
5. Infrastructure - This includes hardware like network including Java, Node.js, and C#. There are over 42 Azure data
devices, servers, storage systems etc. centers located throughout the world, which is the most of
any cloud platform. In addition, Azure plans to add 12 more
6. Security - It implements security mechanisms to protect data centers, bringing the total number of data centers to 54
data stored in cloud, system, and architecture to cloud soon.
users.
B. Amazon Web Services (AWS)
7. Management - It ensures integration and smooth working
of all backend components. Amazon Web Services is a platform that offers flexible and
cost-effective cloud services. AWS is a widely used cloud
platform that provides numerous on-demand activities such
as computational power, database and management,
information and data delivery, and so on to help businesses
develop and flourish. Businesses can use AWS to create a
variety of sophisticated apps. On AWS, organizations of all
sizes and industries may run every potential use case. It was
launched in 2002 and now is among the top cloud service
providers and competitor to Azure and GCP. Azure offers
over 170+ cloud computing services.

Amazon claims that the number of active AWS users tops one
million. Many firms around the world use AWS to create,
deploy, and host applications, whether they are IT
behemoths, startups, governments, or retail groups. Netflix,
Adobe, Coinbase, and others are among them.

C. Google Cloud (GCP)

Figure 2
Google Cloud Platform is a collection of services equipped
with cloud techniques provided by Google that operate on the
IV. ABOUT CLOUD SERVICE USED AND THEIR
same architecture as Google's other major applications such
SERVICE COMPARISON
as YouTube, Gmail, and others. Google Cloud Platform
provides over 100 services. It was launched in 2008.
In this paper, to compare the service and performance we
carried out a study on three of the cloud giants ruling the
For their cloud computing needs, companies such as PayPal,
cloud computing market. Those three are
Twitter, Airbus etc. use the services offered by Google Cloud
• Microsoft Azure (Azure) Platform.
• Amazon Web Services (AWS)
• Google Cloud Platform (GCP) V. ON-DEMAND PRICING COMPARISON AMONG
THE THREE
A. Microsoft Azure (Azure)

Microsoft Azure is a service for cloud computing and a web

platform that allows users to access and administer

270

Authorized licensed use limited to: Beirut Arab University. Downloaded on February 22,2024 at 08:57:05 UTC from IEEE Xplore. Restrictions apply.
2021 International Conference on Technological Advancements and Innovations (ICTAI)

TEST2 1235.11 1026.06 1239.37

TEST3 1240.18 1027.38 1235.45
AVERAGE 1248.13 1025.29 1238.54

Table 2: APACHE BENCHMARK

Apache measures the total system’s performance and relates

to the count of requests that the server can reply to when there
are 1 million requests and 100 of them being competitive.
Through the table and graph, we can see that Azure can
handle more HTTP requests than the other two. GCP comes
at second place in this Apache benchmark test for handling
number of requests that the server can reply to. GCP is giving
a tough competition to Azure in this test case. This is an
important factor to consider for customers who plan to use
VM as a host with a high per second request volume. For such
users Azure is the better option among the three.

Table 1 1500

The table illustrates the on-demand pricing for certain 1000

instances by AWS, Azure and GCP. For general purpose
instance, AWS has the cheapest service at $0.154 per hour 500
followed by GCP at $0.156 per hour. Azure is the most
expensive for General purpose instance among the three at 0
$0.166 per hour. For Compute optimized instance, AWS is TEST1 TEST2 TEST3 AVERAGE
again the cheapest at $0.136 per hour followed by Azure at
$0.169 per hour. GCP is most expensive among the three for AZURE AWS GCP
compute optimized instance at $0.235 per hour. For Memory
Figure 3: APACHE BAR CHART
Optimized instance, GCP is the most expensive at $6.303 per
hour, followed by Azure at $0.252 per hour and again AWS
being the cheapest at $0.202 per hour. For Accelerated Dbench test is a free alternative to the netbench utility
Computing instance, Azure offers the cheapest service at produced by the Samba project. This freeware tests disc
$0.526 per hour and GCP the most expensive at $3.839 per performance by sending requests to the file system. The table
hour. and chart show the comparison between the three cloud
services providers obtained from dbench test. From the table
The overall analysis and comparison of on-demand pricing of we can see that the average of Google Cloud is less than
these cloud service providers show that AWS is the cheapest Azure and AWS. Also, the difference between average of
among most of the instance type and will be pocket friendly Azure and AWS is negligible. So, it won’t matter much if the
for anyone who wishes to use it. user selects either of the two considering the disc
performance.
VI. PERFORMANCE COMPARISON
TEST AZURE AWS GCP

For testing performance VM instances of Ubuntu Linux TEST1 69.826 69.822 63.887
Server 16.04 64 bit were created on AWS, Azure and GCP. TEST2 70.426 70.418 63.5841
TEST3 71.865 71.89 63.6282
Phoronix Test Suite3 was adopted for benchmarking since it
facilitates the evaluation of Linux systems and system AVERAGE 70.7 70.71 63.7
performance under specific situations. Installing this package Table 3: DBENCH BENCHMARK
from the Ubuntu Repository is simple. On both platforms, the
Apache, Dbench, and RAM speed benchmark test processes
(included in the Phoronix Test Suite3) were completed.

TEST AZURE AWS GCP

TEST1 1269.89 1022.43 1240.8

271

Authorized licensed use limited to: Beirut Arab University. Downloaded on February 22,2024 at 08:57:05 UTC from IEEE Xplore. Restrictions apply.
2021 International Conference on Technological Advancements and Innovations (ICTAI)

75
70
65
60
55
TEST1 TEST2 TEST3 AVERAGE

AZURE AWS GCP

Figure 4: DBENCH BENCHMARK BAR GRAPH

Figure 5.2: RAMSPEED FLOATING BENCHMARK

RAM speed is a benchmark test for determining RAM VII. CONCLUSION

efficiency. Copy, Scale, Add, Triad, and Average were the
five test configuration for this. All these five test Even though Google cloud is ahead in the race of cloud
configurations are tested for integer as well as floating point. computing than many cloud providers, it still has a long way
to go before it captures the whole market like it did with its
TEST AZURE AWS GCP
search engine. The difference between average of Azure and
GCP was not much in Apache benchmark test, GCP failed to
ADD 9073.35 9088.03 8756.46
give them tough competition in Dbench and RAM speed
COPY 8837.64 8848.58 6572.52
benchmark.
SCALE 8956.66 8980.09 7790.98
TRIAD 9488.59 9482.88 9400.73 Azure has a simple and intuitive UI for controlling virtual
AVERAGE 9368.71 9099.89 8095.64 resources, but there is no option for VM tweaks. AWS, on the
Table 4.1: RAMSPEED BENCHMARK (INTEGER)
other hand, provides additional facilities for system tuning
and more alternatives for administering Linux VM’s.

When it comes to performance of these cloud services, the

performed tests and their results from testing micro instance
VM’s are very similar for both AWS and Azure, but AWS
manages to excel in both Dbench and RAM speech
benchmarks. Nonetheless, testing benchmark performance of
more sophisticated VM instances should be conducted in
conformity with the virtual hardware representation made
Figure 5.1: RAMSPEED INTEGER BENCHMARK
available to the cloud users. The AWS infrastructure is better
suited for Linux VM’s since it provides more possibilities for
As seen through the chart and table for RAM speed Integer end users to customize, as evidenced by memory and RAM
benchmark, AWS and Azure are in neck-to-neck speed test results.
competition, but with AWS having a better average.
However, GCP is still behind in this benchmark test. REFERENCES
[1] What is Cloud Computing? (n.d.). Amazon Web Services, Inc.
For RAM speed Floating-point benchmarks, AWS remains to Retrieved October 2, 2021, from https://aws.amazon.com/what-is-
hold the first position as well. cloud-computing/
[2] What is cloud computing? A beginner’s guide. (n.d.). Microsoft Azure.
Retrieved October 2, 2021, from https://azure.microsoft.com/en-
So, a user looking for better RAM speed and performance can in/overview/what-is-cloud-computing/#benefits
choose to avail services provided by AWS. [3] Google Cloud. (n.d.). Cloud Computing Services |. Retrieved October
2, 2021, from https://cloud.google.com/
TEST AZURE AWS GCP [4] L. M. Qaisi and I. Aljarah, "A Twitter sentiment analysis for cloud
providers: A case study of Azure vs. AWS," 2016 7th International
Conference on Computer Science and Information Technology (CSIT),
ADD 8888.73 8950.33 8449.1 2016, pp. 1-6, DOI: 10.1109/CSIT.2016.7549473.
[5] B. S. Đorđević, S. P. Jovanović and V. V. Timčenko, "Cloud
COPY 9114.04 9119.1 6548.46 Computing in Amazon and Microsoft Azure platforms: Performance
SCALE 8316.65 8322.93 6646.07 and service comparison," 2014 22nd Telecommunications Forum
Telfor (TELFOR), 2014, pp. 931-934, DOI:
TRIAD 8694.09 8732.45 8520.02 10.1109/TELFOR.2014.7034558.
AVERAGE 8635.46 8781.2 7581.94 [6] A. Alkhatib, A. Al Sabbagh and R. Maraqa, "Pubic Cloud Computing:
Big Three Vendors," 2021 International Conference on Information
Table 4.2: RAMSPEED BENCHMARK (FLOATING) Technology (ICIT), 2021, pp. 230-237, DOI:
10.1109/ICIT52682.2021.9491680.

272

Authorized licensed use limited to: Beirut Arab University. Downloaded on February 22,2024 at 08:57:05 UTC from IEEE Xplore. Restrictions apply.
2021 International Conference on Technological Advancements and Innovations (ICTAI)

[7] Z. Liu, S. Wang, Q. Sun, H. Zou and F. Yang, “Cost-Aware Cloud

Service Request Scheduling for SaaS Providers”, The Computer J., vol.
57, no. 2, 2014, pp. 291-301.
[8] U. Tupakula, V. Varadharajan, “Trust Enhanced Security for Tenant
Transactions in the Cloud Environment”, The Computer J., 2014, pp.
1-16.
[9] W. Zhao, P. Melliar-Smith, L. Moser, “Low Latency Fault Tolerance
System”, The Computer J., vol. 56, 2013, pp. 716-740.
[10] T. Erl, Z. Mahmood and R. Puttini “Cloud Computing: Concepts,
Technology & Architecture”, Prentice-Hall, JCS&T, vol. 13, no. 3,
2013, pp. 63-72
[11] GeeksforGeeks. (n.d.). Architecture of Cloud Computing. Retrieved
October 2, 2021, from https://www.geeksforgeeks.org/architecture-of-
cloud-computing/
[12] D. Armstrong, K. Djemame, “Performance Issues in Clouds: An
Evaluation of Virtual Image Propagation and I/O Paravirtualization”,
The Computer J., vol. 54, no. 6, 2011, pp. 836- 849.
[13] R. Krebsa, C. Momma, Kounevb, “Metrics and techniques for
quantifying performance isolation in cloud environments”, Science of
Computer Programming, vol. 90, 2014, pp. 116–134.
[14] What is Azure and How Does It Work? (n.d.). Simplilearn.Com.
Retrieved October 2, 2021, from
https://www.simplilearn.com/tutorials/azure-tutorial/what-is-
azure#what_is_microsoft_azure
[15] S. (n.d.). What Is AWS(Amazon Web Services): Services,
Applications, Advantages and More. Simplilearn.Com. Retrieved
October 2, 2021, from https://www.simplilearn.com/tutorials/aws-
tutorial/what-is-aws#what_is_aws
[16] Perveez, S. H. (n.d.). What Is Google Cloud Platform?
Simplilearn.Com. Retrieved October 2, 2021, from
https://www.simplilearn.com/google-cloud-platform-article
[17] Y. Jin, Y. Wen, Q. Chen, Z. Zhu, “An Empirical Investigation of the
Impact of Server Virtualization on Energy Efficiency for
GreenDataCenter”, The Computer J., vol. 56, no. 8, 2013, pp. 977- 990.
[18] D. P. Singh, P. Kaushik, M. Jain, V. Tiwari and S. Rajpoot, "Data
Storage Security Issues in Cloud Computing," 2021 International
Conference on Innovative Practices in Technology and Management
(ICIPTM), 2021, pp. 216-220, doi:
10.1109/ICIPTM52218.2021.9388321.

273

Authorized licensed use limited to: Beirut Arab University. Downloaded on February 22,2024 at 08:57:05 UTC from IEEE Xplore. Restrictions apply.