— How to reduce deployment time and cost?
— How to cut the time needed to set up a new Java project?
— How to simplify your development environment?
The answer that comes to mind first is virtualization. Let’s see what K&C’s DevOps consulting team has to say:
In its essence, virtualization allows running of a few operational systems on a single machine, which turned out to be really helpful for testing purposes. But high processor loading, excessive usage of RAM and HD space, and the impossibility of reuse in case of unique configurations makes development processes too slow. Thus, virtualization doesn’t fit for software engineers and stakeholders wanting to deploy their app quickly.
In 2010 HashiCorp introduced Vagrant. As a command line utility for virtualization software, it allows running of commands like [create virtual machine] and generation of complicated configurations. As a result, you can type a single [vagrant up] line in your terminal when you want to, say, run your project with PHP isolated in Ubuntu.
Vagrant made virtualization tasks much simpler for developers, but the approach itself remained too resource-intensive, so in 2013 Solomon Hyckes introduced Docker.
Docker is an open-source platform that was created for rapid deployment and allows:
— setting up the same environment anywhere you need
— release of the application code in less time
— performing quality assurance more quickly
This becomes possible thanks to Docker’s container virtualization platform with processes and utilities managing RAM, HDD, CPU, etc. No matter what app you want to run inside Docker — a NodeJS web app, a Selenium server, a Java application or a Python script — each application will run in isolation inside its Docker container.
And you can create as many containers on a single machine as you need!
In Docker, three main components do all the magic:
— Linux Containers, a technology of virtualization that allows running several isolated OS instances on one host
— Cgroups, a Linux kernel feature that limits, accounts for, and isolates the resource usage (CPU, memory, disk I/O, network, etc.) of a collection of processes
— Linux Namespaces (lightweight process virtualization feature), used to organize isolated places called Docker containers
Containers are similar to directories having everything applications may require to work. Each container is created from an image, a read-only template usually stored in a Docker registry (either private or public).
Dockerfile creating an image with Ruby, Sass, and Gulp installed and then launching a container on top of it:
FROM node:6.1-onbuild # Install gem sass for grunt-contrib-sass RUN apt-get update -qq && apt-get install -y build-essential RUN apt-get install -y ruby RUN gem install sass RUN npm install -g bower gulp RUN bower install --config.interactive=false --allow-root ENV NODE_ENV development # Port 3000 for server # Port 35729 for livereload EXPOSE 3000 35729
Sounds great, but in practice, we usually need to launch a few containers at once to build a complete development or testing environment for an application. What to do then?
Docker’s Compose is a tool introduced for defining and running multi-container applications. To configure your app’s services, you must create a Compose file. Then, using a single [docker-compose up] command, you set up and start all services required for your configuration.
Example of a docker-compose.yml file:
app: build: ./app/ ports: - "3000:3000" - "35729:35729" volumes: - ./app/config:/usr/src/app/config - ./app/modules:/usr/src/app/modules - ./app/public:/usr/src/app/public - ./app/.jshintrc:/usr/src/app/.jshintrc - ./app/.csslintrc:/usr/src/app/.csslintrc environment: MOODLE_URL: http://192.168.99.100:8080 MOODLE_SERVICE: appname_mobile_app ANGULAR_EXPIRATION_HOURS_AUTH: 24 moodle: build: ./dockerfiles/moodle/ ports: - "8080:80" links: - db volumes: - ./dockerfiles/appname/foreground.sh:/etc/apache2/foreground.sh db: image: centurylink/mysql expose: - "3306" environment: MYSQL_DATABASE: moodle MYSQL_USER: moodle MYSQL_PASSWORD: moodle
docker-compose up -d
The above-listed file will run an app that exists in 3 different docker images: app (with a nodeJS application), moodle (with PHP scripts and the Apache web server), and db (with a mySQL database).
Docker allows running any platform with its own configuration on top of your infrastructure without overloading its resources as in the case of virtual machines. It lets you put your environment and configuration into the code and deploy it.
As a result, Docker allows our developers to:
— be as close as possible to production
— have a fast development environment available for interactive use
— run different applications on the same host without any conflicts
— dramatically reduce costs by running many instances on a single server
— bring up new processes via containers within seconds
And no wonder that not only K&C but leading software companies like Google and Facebook eventually turned to using Docker for their development matters as well.