THE ROAD TO DOCKER: A SURVEY

: The interest on conventional technologies is declining with the increasing demand on new technologies. In the virtualization industry, container based technology has become the most powerful technologies in the last couple of years. With the arrival of Docker, implementation of containerization technology has become more simplified and efficient. Unlike other virtualization platform, Docker is an open source software container platform that provides some facilities, which are useful for developers and administrators. Docker has the traits of providing fast and lightweight virtualization on operating system-level, because of which Docker has become popular technology to serve variety of cloud platforms. Development cost and time can be brought down tremendously by simply replacing traditional on-going virtual machine with Docker container. Also with the use of Docker, the cost of re-developing the cloud platform can be reduced to great extent.


I. INTRODUCTION
Docker is an open standard platform for building, exporting, and running applications. It can also be viewed as commandline program with a background daemon running and a set of remote utilities that take a systematic approach to solve common software problems and simplifying user experience of installing, running, publishing, and removing applications. Docker provides facility to separate your applications from your infrastructure to deliver software quickly. A Docker container is a software bucket comprising of all the supporting dependencies necessary to run the software independently. Also the isolation and security feature in Docker allow you to run multiple containers simultaneously on a single host machine. [1] This paper provides insight view on technology of Docker, and we will try to examine how Docker has changed the dynamics of whole virtualization industry. The insight of the paper is systemized as follows. In section 2, we will see what Docker actually is. In section 3, we will focus upon Docker architecture and its components. In section 4, we will compare light weight virtualization using Docker and heavy weight virtualization. Section 5 and 6 will give you a small idea of why to use or not to use Docker container. Finally, in section 7 we will try to compare virtualization with containers following with conclusion and future scope of Docker technology in section 8.

II. DOCKER CONCEPT
Docker is a platform designed to make it easier to create, deploy, and run virtualized application containers on a common operating system (OS), with an ecosystem of allied tools. Initially, Docker was created to work on the Linux platform, but has extended to offer greater support for non-Linux operating platforms, including Mac OSX and Microsoft Windows. Versions of Docker for Amazon Web Services (AWS) and Microsoft Azure are also released. [2] Before Docker, deploying software to different environments required huge efforts. Even though these efforts were encapsulated in virtual machines, a lot of time was spent in managing and deployment of these machines, waiting for them to install and boot, and managing the overhead of resource use they created.
With Docker, image can be pulled down from public or private repositories and is ready to run, consuming fewer resources and contained so that it doesn't interfere with other environments.
User don't need to worry about whether his container will going to be shipped to an Ubuntu machine, a CentOS machine, or any other machine; as long as it has Docker installed on it.

III. DOCKER ARCHITETURE
Docker can be thought of as application based on client server architecture, as shown in Figure 2. Docker on your host machine is divided into two parts-a Server Docker Daemon with a REST API and a Client Docker CLI that talks to the daemon. A REST API is one that uses standard HTTP request types such as GET, POST, and DELETE through which programs can communicate with daemon. The Docker daemon is the core component of Docker architecture which receives instructions from Docker client regarding building, running and shipping the Docker containers. Docker provides facility to run the Docker client and the Docker daemon on same device, or client can connected to the Docker daemon running remotely. Communication between Docker client and Docker daemon is done using a REST API, over a network interface.

A. The Docker Daemon
The Docker daemon (see figure 3) is the hub of your interactions with Docker, which waits for Docker API requests and manages the state of your Docker objects accordingly.

B. The Docker Client
The Docker client (see figure 4) is the simplest component in the Docker architecture, which helps Docker users to communicate with Docker. The Docker client is called when user type commands like docker run or docker pull on his machine. Its job is to communicate with the Docker daemon by sending HTTP requests.

C. Docker Registries
Suppose user has created his own images, and he wants to share them with other users. This is where the concept of the Docker registry comes in. Docker registries provide a platform to store the Docker images. Docker Hub and Docker Cloud are two public registries that can be used by anyone to pull or push images, and by default Docker is designed to search for images on Docker Hub. Many companies set up private registries to store and share their proprietary images internally. You can pull any image using docker pull or docker run command. To push image to your configured registry, use docker push command.

D. Docker Images
In Docker, everything is based on 'Images'. Images act as a read-only template for creating new containers. The platform for creating a new image is a base image. For example, your base image can be Ubuntu 16.04 LTS with an Apache web server and you web application installed on it. This approach of creating a new image is called "committing a change". User can create and run his own Docker image by defining steps in a Dockerfile. Each instruction written in a Dockerfile creates an additional layer in the image, as depicted in Figure 5. When user makes a change in Dockerfile and tries to recreate the Docker image, only those layers are recreated in which changes have been made by that particular user. This is what makes images so lightweight, small, fast and reliable, when compared to other pre-existing virtualization technologies.

E. Docker Containers
At first glance, we can say that Docker container appears to be a lightweight form of virtual machines. Docker container is created using Docker images or we can say container is a runnable instance of a Docker image. Containers hold the complete set of dependencies required by an application to run in more confined way. Using the Docker API or CLI, you can easily create, run, move, stop or remove a container.
Docker containers will always be in one of the four states according to the diagram in Figure 6. [4] Figure 6. The state translation diagram for Docker container [4]

IV. DOCKER V/S VIRTUAL MACHINE
Before we compare Docker with Virtual Machine, let us first see what does the term heavy weight virtualization and light weight virtualization means.
Heavy weight virtualization means each and every virtual machine having its own independent operating system. With the help of virtualization software, you can install any guest operating system such as Windows, Linux and Mac on your host operating system. Virtualization software such as VMware Workstation, KVM, Oracle Virtual Box, and Hyper-V comes under Type-2 Virtualization, which means we cannot directly ran these software's on the system hardware. So the new type of virtualization technology called hypervisor came into the existence which can be installed directly on system hardware. It comes under Type-1 Virtualization. [5] But the problem with these software's are that we need to install whole Guest OS also for doing small experiments like checking network connectivity between two virtual machines using simple ping command, which is sheer wastage of resources.
Nowadays, the focus is moving towards lightweight container based virtualization technology known as Docker as shown in Figure 7. With Docker, application can be placed inside the Docker container, providing Container as a Service (CAAS) platform. These containers containing application are easy to handle and can be placed on any type of platform including cloud. Also these types of light weight technology are fast and consume almost negligible resources. Figure 7. Docker v/s Virtualization [5] In early stages, if we try to compare Docker containers to virtual machines we will see that Docker containers have ability to share single kernel and application library. Also Docker container possess lower system overhead(i.e. computation time, memory etc.), due to which the performance of the application running inside a container is generally better as compared to application running within a virtual machine.
Capability of virtual machines can be enhanced using technologies like Intel's VT-x/EPT and VT-d which is already provided by VMware. An Intel VT-x/EPT and VT-d technology provides same processing power to your guest operating system as your host operating system gets and also increases network performance. Also by using these technologies, virtual machine can achieve ring-1 hardware isolation [6]. Ring works as a protected shield to protect sensitive data and functionality from errors and abnormal behavior by improving fault tolerance and computer security [7].
According to Gupta V (2017), creation and launching time of Docker container is much lesser than that of virtual machine. [5] Also containers consume resources according to the need of particular service or an application. Because of which it is possible to execute more containers on system having not that powerful configuration.

V. ADVANTAGES OF DOCKER CONTAINERS
As we know, Docker has been dominating the DevOps conversation since its arrival in 2013. The main reason for its dominance is the benefits provided by Docker container. In this paper, we seek to highlight some of its best features.

A. Free and Open Source
Generally we all use virtualization software's like VMware etc. to create the virtual machines, but most of these software's require commercial licenses for enterprise use. In contrast, Docker is completely free and open source which can be easily downloaded and installed. Also no one will ask you buy license in order to use you application. [8]

B. Ultra Consistent
While creating virtual machine, sufficient amount of computer resources are required and if resources provided are insufficient virtual machine will not work properly. Whereas in case of Docker, resource consumption is very less because of which higher level of consistency can be achieved. [8]

C. Rapid Deployment
In the past, deployment of new hardware used to take few days. But with the arrival of virtual machines, timeframe was reduced down to minutes. Now the appearance of Docker has reduced deployment to mere seconds. Also the cost of creating and destroying the containers is negligible. [9]

D. Portability
Applications created inside the Docker containers are portable and light weighted. These portable applications can be treated as a single unit hence can be moved easily, without affecting the performance of the containers. [10]

E. Isolation
With Docker, every container created has its own set of resources and each container runs independently without any interfere to other running containers. Docker also ensures easy creation and deletion of your application since each application runs within its own container. Deleting containers won't leave any configuration files on your system. Docker not only also ensures easy creation and deletion of an application but also ensures that each application only uses resources that have been assigned to them. [11] F. Security Using Docker is secure because applications and services that are running inside the containers are completely independent from each other. Which means one container cannot poke into another container. Also as discussed earlier, there is no sharing of resources between containers, and each container has its own set of required resources. That means if something goes wrong with one container, the data available in that container will be affected without affecting the rest of the containers. [11]

VI. DISADVANTAGES OF DOCKER CONTAINERS
Before migrating to Docker, there are some facts or drawbacks which should be kept in mind.

A. Persistent Data Storage is Complicated
In docker, when we shut down a container then all of the data residing inside that container gets disappear forever, unless you save it somewhere else first. Although, Docker provide ways to save your data persistently, such as Docker Data Volumes, but using it seamlessly is still very challenging task.

B. Compatibility with Older Machines
Currently, Docker is compatible only with 64-bit local machines. It does not run on 32-bit machines.

C. Containers don't run at Bare Metal Speed
There is no doubt that Docker containers are capable of consuming resources more efficiently than virtual machines. But performance of containers is not up to that level due to overlay networking.

D. Platform Dependant
Docker was initially designed to run on Linux machine. However, nowadays Docker has started supporting Windows, Mac OS X and many other platforms. To make Docker platform independent we will need additional layer between host operating system and Docker. [12] VII. VIRTUALIZATION V/S CONTAINERS It totally depends on user needs whether he want to go with virtualization technology or with containerization technology. A container environment often provides rapid deployment, greater efficiency and better resource consumption. If you are looking for better isolation then system virtualization may be more relevant option. But with Docker things are changing rapidly.
Gupta V (2017) in Table 1 has compared heavy weight virtualization i.e. Virtual Box against light weight virtualization i.e. Docker on factors like size, memory, storage, installation time and boot time. [5]  According to JC Wang (2015) in Table 2, Container starts within a few seconds with faster speed and capable of loading more services by using less space. [13]

VIII. CONCLUSION AND FUTURE SCOPE
Microservices architecture is a core concept around which Docker is built. This concept has reduced the building and deployment time of Docker containers. Also the demand for resources by a Docker container is very low, and it can achieve higher performance. Docker provides some fine features, which guarantee simplified usability and scalability. However, the Docker container still has some issues regarding security, but tools can be applied to strengthen the base of the container. The main purpose of writing this paper was to give users a brief overview about all the important components of Docker and how Docker technology is different from pre-existing technologies. In future, we will try to look into the other features of Docker like Docker Swarm; a technology for managing a cluster of Docker Engines [15] and NFS Storage; a technology to store and update files on remote computers [16]. We believe, with time Docker technology will get matured and will be deployed more widely.