An Introduction to AWS – EC2
(Elastic Compute Cloud)
Rahul Saini1, Rachna Behl2
1Research Scholar, MRIIRS Faridabad, Haryana, India
2Asstant Professor, MRIIRS Faridabad, Haryana, India
1rahulhbc@yahoo.in , 2rachnabhel.fet@mriu.edu.in
Abstract—Most recent two decades IT is following conven-
tional methodologies for dealing with its framework since the
beginning of Cloud administrations, for example, Amazon
Web Services, Google Cloud Platform, IBM Cloud and a lot all
the more giving on the web foundation administrations. Asso-
ciations have acknowledged an adjustment in their working
model, if an association searches for an extension to expand its
registering power, they just get it online by starting a virtual
machine on the cloud. Virtual machines can be immediately
propelled and shutdown through programming interfaces, of-
fering adaptability to the client rather than customary method-
ologies.
There were times when one was restricted by the boundaries
of a machine e.g.; a data scientist has a large scale of data and
would like to perform some analysis, however, encounters an
error as below while uploading this data.
> data <- read.csv (file=’no_of_bug.csv’)
Error: cannot allocate vector of size 500.0 Mb
The error is the result of the unavailability of the RAM
since the operating system does not have enough RAM avail-
able as there are many solutions available of such problems
following the traditional approach one should upgrade the
RAM of the machine with the restriction to the RAM up-gra-
dation limit to the machine’s compatibility. However, there is
another way by introducing the concept of virtualization in the
could with more RAM and CPU.
This paper explores the concept of Virtual machines on
AWS EC2 as they are called instances with many advantages
few of them includes being highly scalable (one can choose in-
stances with more CPU and RAM etc.) most reliable and scal -
able easy to use for users as pay for what being used also al -
lows a selection of different platform (OS) as an outcome of
this paper will create an understating of the working funda-
mental of AWS EC2.
IndexTerms—Amazon EC2, Cloud Computing, Elastic
Compute Cloud
I. INTRODUCTION
TS impression has been altogether changed since the dis-
tributed computing idea being presented as a pay-more
only as costs arise administrations. To have an application
utilizing Amazon Web Service's (AWS) Elastic Compute
Cloud carries numerous progressions to the improvement,
sending and upkeep forms as these progressions might be
extremely gigantic for certain associations anyway EC2
guarantees expanded adaptability, simplicity of organiza-
tion, versatility and the huge help for the outsider adminis-
trations in contrast with the conventional methodologies.
I
A couple of years back a help administrations association
expected to have and keep up its Incident Management Sys-
tem with an immense venture including programming buys.
As the Cloud processing stage has caught the market so
broadly now association can undoubtedly move their prod-
uct benefits all the more safely to the web.
Right now, will dive into the working key of AWS's
Elastic Compute Cloud (EC2) as the working basics are
fundamentally unique concerning the conventional method -
ologies, for example, inhouse framework the board that is
generally acknowledged anyway these distinctions may
prompt inventive sort of changes and requires various in-
struments to pick up perceivability basic the cloud base
foundation.
National Institute of Standard and Technology defined
Cloud Computing:
“Cloud Computing concept enables on-demand network
access to a shared pool of configurable computing resources
such as (e.g. servers, storage, application, network, and ser-
vices) that can be rapidly conducted and released with mini-
mal management effort or interaction with the service
provider. This cloud model promotes availability and is
composed of five essential characteristics, three service
models and four deployment models.” [3]
II. WHAT IS AMAZON EC2
Foundation as-an administration customers are offered a
wide assorted variety of Cloud assets from numerous, circu -
lated Cloud suppliers (e.g., Amazon EC2 [5], and
RackSpace [6], IBM Smart Cloud[7]) provided at various
hourly cost rates. Besides, comparable Cloud assets might
be valued diversely by each Cloud supplier. Besides, Cloud
buyers may demand heterogeneous arrangements of Cloud
assets that may not be accessible in a solitary Cloud sup-
plier. Along these lines, self-rulingly completing asset as-
signment from numerous and self-intrigued Cloud suppliers
while testing hourly cost rates related to Cloud assets is im-
portant to give proficient (i.e., with low portion costs) asset
allotment administrations to purchasers in a unique way.
This burdens the requirement for the operator worldview.
Operators are self-governing issue solvers that can demon-
strate and work together deftly and self-interestedly among
one another.
A famous methodology in cloud-based administrations is
to permit clients to make and offer virtual pictures with dif-
Proceedings of the International Conference on Research
in Management & Technovation pp. 99–102
DOI: 10.15439/2020KM4
ISSN 2300-5963 ACSIS, Vol. 24
©PTI 2020 99
ferent clients. For instance, a client who has made an inheri-
tance Linux Debian Sarge picture may choose to make this
picture open so that different clients can undoubtedly re-
use it.
Notwithstanding user shared pictures, the cloud specialist
organization may likewise give redone open pictures depen-
dent on the regular needs of their clients (e.g., a Ubuntu
web server picture that has been pre-arranged with MySQL,
PHP and an Apache). This permits the clients to just launch
and start new servers, without the problem of putting in new
programming themselves.
Amazon Elastic Compute Cloud (AWS EC2) provides
scalable computing capacity in the Amazon Web Services
cloud. Using AWS EC2 eliminates the need to invest in
hardware so that one can develop and deploy applications
faster. Amazon EC2 can be used to launch as many or re-
quired virtual machines as per need with security configura-
tion and networking and storage management.
For example, AWS powers the biggest website in the
world which is Netflix example.
•EC2 is one of the most popular of AWS offering
•It mainly consists in the capacity of:
◦Renting virtual machine (EC2)
◦Storing data on virtual devices (EBS)
◦Distribute load across machines (ELB) Elastic
Load Balancer
◦Scaling the services using an auto-scaling
group (ASG)
•Renting a virtual server on cloud is the key to
understand how the cloud works
In this thesis, we are going to learn more about Amazon
Web Services EC2.
EC2 is the main service to provide machines on-demand
and nearly all AWS accounts use EC2 instances to a various
extent knowing EC2 is the key thing to understand how the
cloud works.
Fig 1: Simple Storage Services (S3) Diagram
Amazon Elastic Compute Cloud (Amazon EC2) gives an
adaptable processing limit in the Amazon Web Services
(AWS) cloud. Utilizing Amazon EC2 takes out your need to
put resources into equipment in advance so that you can
create and convey applications quicker. You can utilize
Amazon EC2 to dispatch the same number of or as not
many virtual servers as you need, design security and orga-
nizing, and oversee capacity. Amazon EC2 empowers you
to scale up or down to deal with changes in necessities or
spikes in prominence, lessening your need to gauge traffic.
EC2 permit clients to utilize virtual machines of various
setups according to their prerequisite. It permits different
setup alternatives, mapping of individual servers, different
valuing choices, and so on.
We will talk about these in detail in the AWS Products
area. Following is the diagrammatic portrayal of the design.
It permits clients to store and recover different sorts of in-
formation utilizing API calls. It doesn't contain any figuring
component.
Amazon EC2 (Elastic Compute Cloud) is a web adminis-
tration interface that gives a resizable figure limit in the
AWS cloud. It is intended for engineers to have unlimited
authority over web-scaling and processing assets.
EC2 examples can be resized and the number of cases
scaled up or down according to our prerequisite. These oc-
casions can be propelled in at least one geological areas or
districts, and Availability Zones (AZs). Every area involves
a few AZs in particular areas, associated with low inertness
arranges in a similar locale.
III. IN-DEPTH WITH AMAZON ELATIC COMPUTE CLOUD
A、Classification of Elastic Compute Cloud
Amazon EC2 gives a wide determination of case types
advanced to fit distinctive use cases. Occurrence types in-
volve differing blends of CPU, memory, stockpiling, and
systems administration limit and give you the adaptability to
pick the fitting blend of assets for your applications. Each
occurrence type incorporates at least one example sizes,
permitting you to scale your assets to the prerequisites of
your objective remaining task at hand.
1. Different type of EC2 instances
Amazon has thought of a wide scope of Instances that are
intended to satisfy the wide assortment of necessities of an
Type Description Mnemonic
A1
Good for scale-out workloads,
supported by ARM
a is for ARM processor - or as
light as A1 steak sauce
T-family: T3, T3a, T2
Burstable, good for changing
worloads
t is for tiny or turbo
M family: M6g, M5, M5a, M5n, M4
Balanced, good for consistent
workloads
m is for main or happy medium
Compute Optimized
C-family: C5, C5n, C4 High ratio of compute to memory c is for compute
R-family: R5, R5a, R5n, R4
Good for in-memory databases r is for RAM
X1-family: X1e, X1
Good for full-in memory application
x is for xtreme
High memory
Good for large in-memory
databases
High memory is for high memory
Z1d
Both high compute and high
memory
z is for zippy
P-family: P3, P2
Good for graphics processings and
other GPU uses
p is for pictuers
InF1
Support machine learning
inference applications
inf is for inference
G-family: G4, G3
Accelerate machine learning
inference and graphics-intensive
workloads
g is for graphics
F1
Customizable hardware acceleration
f is FPGA or feel as in hardware
I-family: I3, I3en
SDD-backed, balance of compute
and memory
I is for IOPS
D2 Highest disk ratio
d is for dense
H1
HDD-backed, balance of compute
and memory
H is for HDD
General Purpose
Memory Optimized
Accelerated Optimized
Storage Optimized
Fig 2: Different types of EC2 instances
100 PROCEEDINGS OF ICRMAT. NAGPUR, 2020
association. We have recorded distinctive AWS EC2 occur-
rence Types underneath with the highlights. How about we
find a workable pace each EC2 example type in a superior
way.
This diagram shows a brisk rundown of the EC2 in-
stances, below there is a short description of each category
of the EC2 instances.
2. Issues with EC2 performace
I. EC2 instances can be optimized with effective utiliza-
tion and effective services used however in how the re-
sources are delivered and utilized, EC2 instances can be op -
timized.
II. This is the original EC2 architecture, you see every-
thing is on the server, Network, Storage, Management, se-
curity, and monitoring which means it is taking the re-
sources from the server.
Fig 3: EC2 Architecture
That means that we can utilize the resources in such a
way where we can make some of the instance free.
“Launching new instance and running tests in parallel is
easy however there is no substitute for measuring the per-
formance of your full application.”
III. Choosing a right AMI plays an important role in the
performance of the EC2 instances since the AMI includes
such information about the launching environment of the in-
stance that may disrupt the performance of your EC2 in-
stance such as template of the root volume which includes
(operating system, application), launching permission that
control which AWS account to be used and block device
mapping that specifies volume to attached to the instance.
Some of the tips to choose the right AMI and OS, choose
the latest OS level your tool or application supports
The kernel should be at 3.10 or higher
•As much as 40% performance management
•Should not be using a 2.6 or older kernel
Minimum recommended OS*
•The most recent version of Amazon Linux 2 or
Amazon Linux AMI
•Ubuntu version 16.04 or latest LTS release
provided by AWS
•Red Hat Enterprise Linux version 7.4
•CentOS 7 version 1708_11
•SUSE Linux Enterprise Server 12 SP2
•FreeBSD 11.1 or later (does not support F1
instances)
IV. Processor state control also important to improve the
performance of the EC2 instances there are two types of
state control instances
C-state - entering the deeper idle state, allows active
cores to achieve higher clock frequency, but the deeper idle
state requires more time it exit may not be appropriate for
latency-sensitive workloads, Windows: no option to control
c-state
P-state - controls the CPU’s ability to change frequency,
including enabling or disabling Turbo boost
V. Amazon Elastic Compute Cloud (AWS EC2) provides
scalable computing capacity in the Amazon Web Services
cloud. Using AWS EC2 eliminates the need to invest in
hardware so that one can develop and deploy applications
faster. Amazon EC2 can be used to launch as many or re-
quired virtual machines as per need with security configura-
tion and networking and storage management.
3. Challenges with EC2 instances
We have studied EC2 instances performance depends
upon multiple factors in the cloud architecture. Our objec-
tive is to identify the factors which may impact the perfor -
mance of the instances.
•Storage Latency
The problem that we faced more frequently with the EC2
instances such as Storage latency, In AWS EC2 there are
two types of storage volume in Elastic Block Storage (EBS)
as Standard volumes and Provisioned volume the basic dif -
ference between these volumes is provisioned volume per-
form much faster than the standard volume for e.g: an oper -
ation demands a throughput at the rate of 10,000 IPOS (in-
put/output per second), considering the several bands only
provisioned volume can get you that.
Increasing the IPOS may lead to other issues such as EBS
do not support the given rate of the IPOS, the tasks many
begin to queueing up or application may get fail.
To handle such issues we should keep track of IPOS,
there is a feature in AWS called VolumeQueueLength
which tracks the number of IOPS operations in the queue.
•EM2 Memory Leaks
There is a high probability that instances may run out of
memory. The chances are high as EC2 instances may not in-
clude the swap volumes resulted unable to free the memory
with applications at the time it is being allocated to the new
application. Such cases can be considered a good example
during the heavy traffic on the application, the instances
may freeze the memory while the application is going
through with the peak load, therefore, the purpose of EC2
instances are defeated in the first place.
Such scenarios are difficult to handle however adding up
more EC2 instances may get beneficial to optimize the per-
formance of your application, however, on the other hand,
killing these processes may help however increases the risk
or a failover of your application as a vital process may get
killed. Adding more EC2 instances could be expensive
however this is one of the better ways of doing it.
RAHUL SAINI, RACHNA BEHL: AN INTRODUCTION TO AWS—EC2 101
•Idle EC2 Instances
At launch all the EC2 instances are idle or the scenario
could be anything for e.g: an instance having an average
CPU utilization less than 2% in last week or the network
I/O is less than 10 MB last week so defining a threshold of
an instance to consider an idle instance depends upon the
configuration of the instance or the need of your organiza-
tion.
There is a tool that has been launched with the new fea-
tures that provide the recommendations to optimize the EC2
instances. The main purpose of the tool to monitor the uti-
lization of your EC2 instances upon defining the thresholds
of the instances any breaches to the set values such as CPU
utilization is lower than 1%, the toll suggests to terminate
the instance since this is no longer active.
IV. CONCLUSION
This paper elaborates an overview of Amazon EC2 in-
stances which cover the types of EC2 instances as described
in Fig:2 and major performance issues with EC2 instaces
are effective utilization, creating and using an EC2 in-
stances AMI plays vital role in performance of an instances.
Challenges with EC2 instances is explained, a practi-
tioner to consider which contributes effectivtly to improve
the performance of EC2 instacnes.
REFERENCES
[1] Alexis Lê-Quô, Mike Fiedler, and Carlo Cabanilla, “AWS EC2
Performance Problems,” Datadog Inc., March 2013.
[2] Marco Balduzzi, Jonas Zaddach, Davide Bazartti, Engin Kirda, “A
Security Analysis of Amazon’s Elastic Compute Cloud Services”,
EURECOM: 2012.
[3] M. SEAGRAVE, “How cloud computing shaping the IT job market,”
Guardian, https://careers.guardian.co.uk/careers-blog/it-job-market-
cloud-computing, accesses, April 2012.
[4] B. Simpson, et al, “Title of paper goes here if known,” unpublished.
[5] Amazon elastic compute cloud, May 2011.
http://aws.amazon.com/security
[6] Cloud computing, cloud hosting and online storage by rackspace
hosting, May 2011. http://www.rackspace.com/cloud/
[7] Ibm smartcloud, May 2011.
[8] https: //www.ibm.com/cloud-computing/us/en/#!iaas
[9] https://aws.amazon.com/about-aws/
[10] https://www.visual-paradigm.com/guide/cloud-services-
architecture/what-is-aws-architecture/
[11] Cloud computing, cloud hosting and online storage by rackspace
hosting, May 2011. http://www.rackspace.com/cloud/
[12] M. Young, The Technical Writer’s Handbook, Mill Valley, CA:
University Science, 1989.
102 PROCEEDINGS OF ICRMAT. NAGPUR, 2020
