Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)

DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

Energy Efficient Next-Gen of Virtualization for

Cloud-native Applications in Modern Data Centres
Mr.BharaniDharan.G1 Dr.Jayalakshmi.S 2
1 2
Research Scholar, Dept. Of Computer Science, VISTAS, Professor, Dept. Of Computer Applications, VISTAS,
Pallavaram, Chennai. Pallavaram, Chennai.
Email: bharanmca@gmail.com Email: jai.scs@velsuniv.ac.in

Abstract—In the new software-dri ven worl d this is the with the same operating systems and runtime co mponents.
need of the hour to achieve success by accelerating Containerizat ion is evolving and provisioning major impact in
development, deli very of applicati ons and services rapi dl y Modern computing technologies. By means of containers,
that make customers happy and business competiti ve. To developers can encapsulate all co mponents, binaries, libraries
achieve faster deli very of resources and applications to and make it as separate micro-services of an applicat ion with
strengthen the business with more transformati ons continuous integration exercise process to make the
containers have been introduced. At the existing scenario, application more ag ile with resilient. Applications in the
the hyper -converged data centre uses the concept of microservice approach consist of many decoupled services
virtualizati on which creates software abstracti on of the independent of each other. Each of these services performs a
underlyi ng hardware using Hypervisor software that specific task that is developed and deployed independently [2].
enables to execute several Virtual Machines (VMs) wi th
dissimilar Operating System (OS) fl avours. B ut Docker is also playing an emerging role in container-based
virtualizati on is heavyweight and it may take more ti me t o cloud autonomic data centre. Scalable container service has
boot. The proposed containerization and docker in hybri d been used by most cloud service providers like IBM, Amazon,
cloud composable data centre is lightweight supports Google, Alibaba. Docker containers can easily be deployed
Operating System (OS) level virtualization that isolates into the cloud-based environment [3].
resources, libraries and other bi naries bundled into a The unique approach has been utilized by container-based
single package for agile modern scalable workloads. This virtualizat ion illustrate about the virtualization layer which is
paper focuses on the container-based virtualization to available on the top and the application runs within the OS. It
shape the future cloud-based modern data centre for has to be noted here as default host OS is considered at the
supporti ng the micro-services based applications for faster base [4]. In container-based Virtualization, a single OS will
depl oyment to Developer Operations (DevOps) teams in IT take care of all hardware calls.
wi th resiliency, High Availability and better resource In Containers, the layer of virtualizat ion has provided a file
management with energy savings. system and similarly, the layer o f kernel service separation has
assisted to isolate resource from all VMs to make containers
Keywords-Hypervisor, Container, Docker, Virtualization, look like an individual server [5]. Th is paper is a review of the
Agile, High Availability, Hyper-Converged, DevOps, Micro- technology of docker and analyses its performance by a
services. systematic literature rev iew. The wo rk is organised as follow.
Next section discusses the literate review of existing VMs and
the Next-Gen v irtualisation is container-based. In Sect ion III,
I. INTRODUCTION the benefit of Container-based virtualization whereas Section
deals with the co mponents and significants of the docker
In the past decade, the computing paradigm has a rapid shift container. In section V, the architecture of docker is illustrated
fro m on-premise trad itional infrastructure to on-demand and similarly in section VI microservice arch itecture and
Hybrid cloud Infrastructure. Most of the IT organisations and challenges. In section VII the arrangements of the container by
enterprises moved to cloud for the development of core Kubernetes And Docker Swarm. Section VIII illustrates the
business. Cloud Serv ice Providers (CSP) has been providing comparison of performance evaluation in proposed
flexib le, scalable and elastic infrastructure of IT in term of architecture with existing VMs arch itecture. Section IX
virtualizat ion technology with VMs [1]. The usage of VMs concludes that software-defined DC may rely on container and
may incur energy and wasting the computing resources by VM p lacement mechanisms to support next-generation
running the same operations and content with various guest applications.
operating systems. To avoid this inefficiency and energy
consumption, the container technology has been introduced to
increase resource usage efficiency for business agility and
improvement in deploy ment by sharing the same infrastructure

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 205

 Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)
DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

II. LITERATURE REVIEW TABLE. I: Ideal d ifferences between Virtual Machines and
Containers
In Modern clouds, Virtualization is one of the vibrant concepts S.No CHARACTE VMs CONTAINERS
for sharing physical resources among various users [6]. . RISITICS
Virtualizat ion based on hypervisor helps the data centre 01 Abstraction Hardware-level OS-level
resources can be utilised efficiently by means of consolidation abstraction abstraction
to a single system and even assist for fault tolerance and 02 Number of Multiple kernels Single kernel
replication with other geographical areas [7]. kernels for multiple for various
Virtualizat ion is widely classified into two categories such as VM’s on a containers on a
Full v irtualization and Paravirtualizat ion. In the process of physical physical
Full virtualizat ion has created virtual resources such as machine machine.
storage, processor and network to run various flavours of 03 Booting time Takes a few Takes a few
guest OS on a single physical machine which are unaware that minutes seconds
they are existing in virtualised location [8]. Most importantly 04 Live Migration Takes more time Shorter time
all operating systems think as it runs on bare-metal. Full Time because of because of
Virtualisation is imp lemented through VM ware. It is running heavy migrating
important to note that windows Hyper-V use Microsoft kernel applications specific OS
and ESXi server by Linu x Kernel. In Para-Virtualisation to components
satisfy the necessary resource requirements, the applicat ions instead of the
of guest are acco mplished in quarantined fields wh ich have whole OS
identified as guests and instruct command straight to the OS of 05 Predictability Though post Predictability
the host. The Para-Virtualisation can be used with the Zen in Live- copy offers due to smaller
family. Virtualization is heavyweight and can cause high Migration better prediction memory
resource overheads. but not footprint.
One of the significant factor done through microservice satisfactorily
virtualizat ion is to develop the cloud application performance [12]
whereas there are several researchers involved in analyzing
various virtualizat ion technique performance [9]. The alternate
emerging technology to virtualization is containerizat ion that
assists in obtaining popularity in VMs because of its high
scalability, high performance and light weighted. The Barik et
al. has made a co mparative analysis over the performance of
VMs and container with several simu lations [10]. However,
the architecture of container differs fro m VMs and the main
distinguish is about sharing of host OS can be done by
container but in virtualization it can be done at kernel level.
Therefore, the requirement of container is guest process which
has been compatible with host kernel is shown in figure I.
Thus, container is compared to VMs in wh ich every VM has
its individual guest OS whereas the container reduced its
overhead by introducing dissimilar guest OS to various
container process. This get resulted in requirement of less
memory, reduced infrastructure cost with high performance.
The other Next -Gen v irtualisation is container-based or OS-
level based virtualisation is an alternate to hypervisor-based
virtualisation effectively reduces resource overheads and
improves resource utilisation with faster deployment in FIGURE.I: Virtual Machine vs. Container [13]
modern data centres for heterogeneous modern workloads is
illustrated in Table I. In containerisation, each separate case is
said to be Container which executes on the top of a shared OS III.BENEFITS OF OS-LEVEL BASED OR
along with proper and required isolation. In Containerisation CONTAINER VIRTUALISATION
the virtual objects are restricted to global kernel resources
which result in utilizing resources like networking, CPU and
memo ry for cost-efficient management, scalable and Co mpared to conventional Hypervisor based virtualization,
effectiveness efficient over cloud infrastructure [11]. containerizat ion improves performance, energy and efficiency
because of additional resource required for each OS is
avoided. The Container-based virtualisation fits best when a

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 206

 Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)
DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

single OS is required. Single Kernel is used to execute several  Utilization of resources: The physical servers can
cases of an OS. In the OS level, virtualisation doesn’t able to execute more containers than VMs which can
duplicate its functionality. result in high resource utilization. Ho wever, the
In Container-based virtualisation instances are very smaller containers are thin, effective and portable that can
and it is faster and easier for generating or mig rating. Cloud able to execute on the physical host. Containers run
Service Providers (CSP) have involved in containers due to as an isolated process and it contains application code
the same hardware environment in wh ich more instances of a with dependencies and also shares a kernel with
container can be implemented. various containers in the user’s host OS space [17].

 Challenges faced in the use of containers:  Exact fit for Microservices architecture: Containers
support microservices architecture as every
One of the major disadvantages present in container-based microservices that can be implemented with no
virtualizat ion is security. However, the containers can able to disturbance with other microservices. Containers are
share the components and kernel o f the host OS. It is to be appropriate to service arrangement by means of
noted here that Virtual Machines shares only the hypervisor agility, separation and easy deployment with new
make less prone to attack [14]. versions.
In containers, the host OS has created a single point situation
on failures of host OS may affect any containers.  Portability: Docker can embed any kind of
application needs over a container is portable across
various platforms. The application of distributed can
IV. DOCKER CONTAINERS able to create, imp lement and execute through
containers. Inside the containers deployment,
orchestration can be done so that application
The name of the organizat ion is Docker that creates software developers can execute a similar applicat ion on VMs
named as Moby which is an open-source project whereas the or in the cloud.
Docker gets executed on Windows and Linu x. The usage of a
container is quite simp le to generate, imp lement and execute  Minimal resource utilization: Docker can allocate
applications designed with Docker tool. Docker helps the limited resources to any kind of processes by LINUX
developers to get a clear view of the stack easily. control groups. It confirms that a single process
doesn’t consume more co mputer resources on
underfed other processes.
A. The different parts in Docker are:

 Docker Engine: It performs as a program wh ich V. DOCKER ARCHITECTURE

generates and executes Docker Container fro m the
Docker image itself.
 Docker images: It is a file system and organised with A. Docker Engine:
several layers and every layer consists of file for that
layer wh ich is immutable. Docker images are used as It is performed as an application of client-server and it gets
a Docker container snapshot. comprised into three parts namely
 Docker Files: It perfo rms as a text document wh ich
consists of guidelines to co mpose an image that can
be recognised using build engine. Docker file also
 Daemon Process: Th is is a server process which runs
as a background process that executes continuously
defined the scenario within the container. Inside the
to process any commands by constantly listening to
container mapping volu mes, the files get copied and
REST API.
access to resources may happen.

B. Significant advantages of Container Dockers  REST API: REST API can able to access via HTTP
are: client and to communicate with Docker daemon.

 Fast and lightweight Docker: Containers have

performed as a lightweight and boots fast when  Client: It is performed as a Co mmand Line Interface
compared to VMs because VM makes use of the (CLI).
entire OS to start. It is to be noted that each VM runs
full OS instance to consume resources [15][16].

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 207

 Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)
DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

B.Architecture of Docker Client-Server: The Docker client with the build, pull and run has been
communicated to the daemon of docker that perform in
The architecture of Docker system is majorly involved with generating better lift in the build ing, distributing and executing
client, registry and DOCKER-HOST (Daemon). the docker container. However, the execution is done either on
a similar system fro m docker client and docker daemon or it
 Docker Client: Through the Docker client users can can able to connect docker client through remote of docker
interact with Docker Daemon using the co mmands in daemon. Therefore, the co mmunication of docker client and
CLI through Docker API interface. However, the daemon has used UNIX network interface by REST and API
Docker objects can be created by the command that is shown in figure II.
gets executed by Docker. Hence, the Multiple Docker
Daemons have been accessed by the Docker client.
VI. EVOLUTION OF MICROSERVICES
 Docker Registries: Docker registry is a place where
the images are created using Docker daemon wh ich In a monolithic applicat ion where the co mponents of the
application, interface, data access code has been strongly
gets shared in a location. Moreover, the available
coupled as a single program. Most of the application is still
public reg istries are Docker hubs that can be accessed
fro m all users. Hence, it is to be noted that the based on monolithic because it is easy to imp lement, test and
deploy.
Docker hub can also be utilized fo r configuring
private registry. Thus, the Docker PULL is utilized to
retrieve an image fro m the registry wh ich gets Features of Microservices:
configured and similarly Docker PUSH is used to  Agile delivery: Services get decomposed into
store an image in the registry. logically modular in which the independent
microservices assist in to perform with Agile
 Docker daemon: It performs as a server process delivery, easily fits over DevOps model and faster
which is determined and runs in the background time to market.
looks on to REST API for the inco ming request and  High performance and high availability:
to execute commands. API interface can be listened Containerized microservices has been leveraged to
by daemon using File Descriptor, UNIX, high performance and availab ility. The request of
Transmission Control Protocol (TCP) sockets. microservices with asynchronous nature assists in
improving the performance.
 Docker Objects: The objects represent services,  Resiliency and fault tolerance: Container eco-system
containers, images and storage. offers features such as clustering, load balancing,
circuit breaking and others to offer high resiliency
 Images: To build an image, Docker file is used and and fault tolerance for the microservices. Thus, the
the file system is a read-only that consists of design is provided for with ag ile functionality
commands for creating a container to execute an degradation.
application. Thus, the Docker images can be utilized
for imp lementing an applicat ion overproduction or
test environment. Challenges of Monolithic services:
 Difficult to test different modules of an application
independently.
 Change in a piece of code in a particular module,
deployment of the entire service will take a long
time.
 Monolithic services have a single point of failure
because bug present in any modules can down the
complete service.

To avoid all the pitfalls of Monolithic arch itecture,

microservices came to existence.
In the architecture of Microservices has an application wh ich
gets divided into several services which are independent to the
database. It gives developers a lot of sophistication to choose
the individual modules to work. These services are also being
scaled based on their requirement and the services of testing
FIGURE.II: Docker Architecture [18]
independent are quite simp le that brings modularity is
illustrated in Table II.

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 208

 Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)
DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

TABLE.II: Co mparison of Monolithic Architecture and

Microservices VII. DOCKER CONTAINER O RCHESTRATION USING
KUBERNETES AND DOCKER SWARM
FEATURES MONOLITHIC MICROS ERVICES According to the Docker container, both Kubernetes and
ARCHITECTURE ARCHITECTURE Docker swarm is the orches trating tools. Kubernetes is open-
Maintenance Maintenance is Maintenance is easy source software for containers arrangements. In Greek,
complex of because they are Kubernetes is also represented as K8s which illustrate the
coherency of independent. mean ing as a pilot. However, Kubernetes is the ship captain
modules. with all executing containers present inside [19]. Therefore,
Deployment Containers Deploying individual Kubernetes container management has responsibilities in
deployment is services is easier. involving container implementation, load balance and scaling.
difficult because of Thus, it performs as an open-source platform fo r managing
its expansive nature. containerized applications and to automate deploy ments [20].
Testing Testing the entire Individual Kubernetes is for organizing and running multiple connected
modules that are components in containers in rack server based Performance-Opt imized
coupled is a Microservices are Datacentres (PODS). According to the arrangement, the
cumbersome task. easier to test. container Kubernetes is in the de-facto standard whereas
Scalability Harder to scale and It can be scaled on Kubernetes has supported various modern workloads namely
no flexibility demand stately, stateless and processed workloads with boundless
Start-up time More time to boot Due to the smaller swift [21]. Kubernetes helps us to create ingress routing and to
size of individual run and execute stately services. It will also help to maintain
services, start-up time secret and manage passwords. Kubernetes can be useful in
is less. deploying a large cluster, managing a large nu mber of
Technology A monolithic Adopt to any new containers to reliable and quick response time.
application is written technologies because
in a single language of each Microservice
with a single use the appropriate A. Kubernetes Features:
database. database that fits the
needs  Load balancing with service Discovery: Kubernetes
automatically maintains all networking
communicat ion by assigning IP addresses in all
containers and a DNS name to container set for
maintaining load balance within the clusters is
illustrated in Table 3.

 Auto Bin-Packing: Based on the requirements

Kubernetes automatically schedules and pack the
applications within the containers. It provides best-
effort workloads by ensuring complete utilizat ion of
resources and saves unused resources [22].

 Storage Automation: With the aid of Kubernetes

mounting, the storage system will be done based on
user choice. User can choose for public cloud
FIGURE. III: Microservices provider storage namely AWS or local storage or by
shared network storage such as Network File System
Each M icroservices is for the function of specific business and (NFS).
it gets suitable in operations of particular business function is
shown in figure III. Ho wever, the architecture of
 Automatic Rollbacks: Kubernetes automatically roll
Microservices has provided web based advancement for better back the changes for the user if so mething goes
flexib ility and agile to support and manage code base. wrong with the configuration or not proper updates
Therefore, the Docker is performed as an organizer and eye-
installed to an application.
opener to container and microservices oriented application
deployment.

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 209

 Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)
DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

 Self Healing: During the execution, if the containers  Kublet: One of the significant parts of the slave node
fail, Kubernetes automatically restart the containers is Container Engine (CE) and it will be essential and
and kill those containers that do not respond to make it possible for executing containers. Kubernetes
custom health checks [23]. can talk to Kublet for executing container that
executes on worker nodes.
B. KUBERNETES ARCHITECTURE
 Pods: Kubernetes can take care of any hosts present
Kubernetes architecture mainly comprises of in a cluster and can join any hosts available over a
cluster to run container is known as POD. POD is
1.Master Node: mainly used for redundancy, scalability [25]. Group
of POD’s that can form an applicat ion and similarly
Kubernetes cluster can be managed by the Master Node and it storage is referenced to the POD wh ich can be inside
is the waypoint to the entire ad min istrative task. The or outside the cluster to the POD. Thus, the Kube-
controlling point is hosted by the Master Node is available for proxy has performed as a network pro xy that can
the whole environment is shown in figure IV. So me of the execute on every node over the cluster.
services in Master Node are:

 Control Manager: It manages the process and the

service that executes together on Kubernetes Master.

 API Server: This is to showcase the functionality to

the users. Using kubcctl co mmand the user can
manage Kubernetes and also can talk to the API
server.

 Scheduler: In some environment whether the

Kubernetes pods or implements have been scheduled
properly will be ensured by the scheduler.

 ETCD: It is the back end database with key-value

pair gets stored in the database [23]. FIGURE.V: Kubernetes Architecture-Slave Node [26]

3. Kubernetes Cluster:

It contains a set of node mach ines for running

containerized applications. A cluster contains a slave
node and a master node. Running Kubernetes has
shown running a cluster. Deploying programs on to
the cluster intelligently handle the distributed work to
the separate nodes. Kubernetes can able to supports
clusters up to 5000 nodes.

C. DOCKER SWARM

A swarm acts as a machine crew wh ich can be either physical

or VMs that assist in executing the application of Docker and
it gets configured for merging in a cluster is illustrated over
FIGURE.IV: Kubernetes Architecture (Master Node) Table 3. The Docker swarm has assisted user for managing
several containers generally which has been imp lemented over
2. Worker-Slave Nodes: various host machines. Swarm directors are the root machines
in a swarm that can process the direction or allo w d ifferent
mach ines to join the swarm as workers or labourers. The IT
Apart fro m the Master Node, there are also slave administrators, developers can be easily managing a cluster of
nodes [24]. If the architecture is large, mo re worker Docker nodes as a single Virtual machine is shown in figure
nodes will be there is shown in figure V. VI.

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 210

 Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)
DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

VIII. RESULT AND DISCUSSION

Docker Swarm Features: In this session, the implementation strategy of elastic
resource allocation depends upon the performance measure
whereas the proposed Docker containerization DC has
 Autoload balancing: There is autoload balancing
determined the suitability to meet the requirement in
present in the environment, and it can able to script various workload types using cloudStack4.5. Hence, th is
that into how it can be written out and structure the
proposal has considered with standard DC in a co mbination
Swarm environment.
of 15/15 as CPU/Memory modules using 3.4 GHz and
 Decentralized access: Swarm makes it very simple to capacity as 24 GB capacities correspondingly over
the teams for accessing and managing the CentOS6.9 for co mparison of Docker containerization
environment architecture of DC among existing VM architectured DC.
 High scalability: Load balancing has converted the
Swarm environ ment into a highly scalable
infrastructure. Resource Energy consumption

Types of nodes in the swarm cluster: The ratio of total energy consumption and total interval
numbers is said to be average energy consumption is
formulated in equation 1. In addition, energy consumption is
 Manager Nodes: It will handle the cluster
management tasks such as scheduling services, formulated in equation 2.
maintaining the state of the cluster and serving ∑
swarming mode through HTTP API endpoints. (1)

 Worker nodes: There are instances of Docker engine

particularly to execute containers. I do not involve in ( ) (2)
raft distributed state like scheduling decisions. It
should be noted that no worker node without at least Where,
one manager node. K = the coefficient of idle power consumption = 0.7
Pmax = Peak power
u = Utilization of CPU

The power consumed fro m the CPU during workload process

and ideal are evaluated for all type of DC infrastructure
whereas the proposed architecture DC consumed less power
compared than that of VM in frastructure DC shown in figure
VII due to high agility and automatically schedules of the
applications within the containers by Kubernetes.

FIGURE. VI: Swarm Cluster [27]

This proposed architecture Orchestration of docker

containerizat ion has implemented with the modern DC to
discover better energy efficiency and high agility due to
FIGURE. VII: Comparison of energy consumption
automating deployment of Kubernetes and auto load balancing
of docker swarm. The proposed architecture of docker
conterinization is compared with existing VMs.

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 211

 Proceedings of the Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)
DVD Part Number:CFP20OSV-DVD; ISBN: 978-1-7281-5463-3

[10] R. K. Barik, R. K. Len ka, K. R. Rao and D. Ghose,

“Performanceanalysis of virtual mach ines and containers in
IX. CONCLUSION AND FUTURE WORK cloud computing,” 2016International Conference on
Co mputing, Co mmunicat ion andAutomation (ICCCA), Noida,
2016, pp. 1204-1210.
To cope up with executing modern applications, the Software-
Driven data centres are relied on orchestration using [11] JunzoWatadaet. al., “Emerging Trends, Techniques and
open issues of Containerizat ion, DOI: 10.1109/ACCESS.
containers, Dockers and Kubernetes to provision the resources
2019-2945930, 2019.
to the cloud-native, modern workloads or the Dev-ops teams
with agility by means of on-demand. This proposed [12] PetterSvard et al., “Principles and Performance
Characteristics of Algorith ms for LIVE VM Migrat ion. ACM
architecture has evaluated in term of power consumption
SIGOPS Operating Systems Review, 49(1): 142-155, 2015.
which illustrates less energy consumption than VMs
architecture. The Modern infrastructure also provides auto- [13] https://blog.netapp.com/blogs/containers -VS-vms/
[14] KinnaryJangla, Accelerating Develop ment Velocity using
scaling with proactive and predict ive modes, load balancing
Docker: Docker across Microservices, IEEE, ISSN: 978-1-
using server consolidations and VM placement mechanis ms to
support the next-generation applications which will help for 4842-3936-0, 2018.
[15] Docker leads the container technology charge in cloud-
the enterprise business growth and expansion that enhanced to
http://serachcloudcomputing.techtarget.com/feature/Docker-
generate the application deployment still mo re incred ible,
faster and automated. In future works , the study will be leads-the container-technology-charge-in-cloud.
[16] containers-Not Virtual Machine-Are the future cloud-
conducted on Ansible container and playbooks can be used.
http://www.linu xjournal.co m/content/conatainers%E2%80%9
4 not-virtual-machines %E2%80%94 are-future-
cloud?page=0,1.
REFERENCES [17] M icroservices Architecture, Containers and Docker.
http://www.ib m.co m/developerworks/community/blogs/lba56f
e3-432f-alab-
[1] AkkaritSangpetchet. al., “Thoth: Automatic Resource 58ba3910b073/entry/microservices_architecture_containers_a
Management with mach ine Learning for Container-based nd_docker?lang=eng.
cloud platform”, IEEE, Proceedings of 7th International [18] KinnaryJangla, Accelerating Develop ment Velocity using
Conference on Cloud Co mputing and Services Science, 2017, Docker: Docker across Microservices, IEEE, ISSN: 978-1-
ISSN: 978-989-758-243-1. 4842-3936-0, 2018.
[2] Z. Xiao, I. Wijegunaratne and X. Qiang, ”Reflections on [19] Docu mentation of “PaaS and Container Clouds” by John
SOA andMicroservices,” 2016 4th International Conference Refrano, IBM,NY University, 2015.
on EnterpriseSystems (ES), Melbourne, Australia, 2016, pp. [20] Publicat ion of “Kuberenetes Fundamentals” by the Linu x
60-67. Foundation training, 2017.
[3] Bui, T. (2015). Analysis of docker security.arXiv preprint [21] “Bringing Devops for Netwo rk with Ansible” by Redhat
arXiv:1501.02967.
Enterprise.
[4] Sachchidanand Singh et. al., “Conatainers and Docker: [22] Basit Mustafa, Taow. James Lee, Stefan Thorpe
Emerging Roles and Future of Cloud Technology, IEEE, “Kubernetes from the ground up, deply and scale performance
2016, ISSN: 978-1-5090-2399-8/16.
and reliable containerized applications with Kubernetes” by
[5] Virtualization performance and Container-based Level Up Kubernetes Program, 2018.
Virtualizat ion, [23] Article on “What is Docker Container?” by Saurabh
http://searchservervirtualization.techtarget.com/tip/virtualizati
Kulshrestha.
on-performance-and-container-based-virtualization. [24] Article on “How does Kubernetes Networking work” by
[6] David Bernstein, Containers and Cloud: Fro m LXC to Level up Programming.
docker to Kubernetes, IEEE Cloud Co mputing, 1(3): 81-84,
[25] Arundel, Khare, Saito, lee and Carol Hsu “Devops:
2014. Puppet, Docker and Kubernetes -learning path” by Thomas
[7]Min xian Xu et. al., “A survey on load balancing algorith ms Uphill, Packet publications, First Edition, 2017.
for VM Placement in cloud computing, Concurrency and
[26] Jay Shah and DushyantDubaria, “Bu ilding Modern
computation: Practice and Experience. 29(12): e4123, 2017. Clouds: using Docker, Kubernetes and Google Platform,
[8] Rich Uh lig et al., Intel Virtualization Technology IEEE, ISSN: 978-1-7281-0554-3/19, 2019.
Computer, 38(5):48-56, 2015.
[27] Nikh il Marathe, Ankita Gandhi, Jaimeel M Shah, “
[9] H. Khazaei, C. Barna, N. Beigi-Mohammad i and M. Docker swarm and Kubernetes in cloud computing
Litoiu, “EfficiencyAnalysis of Provisioning Microservices,” Environment”, IEEE Xplore, ISSN: 978-1-5386-9439-8, 2019.
2016 IEEE InternationalConference on Cloud Co mputing
[28] Dmitry Paunin, “The best architecture with Docker and
Technology and Science (CloudCo m),Lu xembourg City, Kubernetes-myth or reality.
2016, pp. 261-268.

978-1-7281-5464-0/20/$31.00 ©2020 IEEE 212