Photo by Cameron Venti
Docker, a subset of the Moby project, is a software framework for building, running, and managing containers on servers and the cloud. The term "docker" may refer to either the tools (the commands and a daemon) or to the Dockerfile file format.
It used to be that when you wanted to run a web application, you bought a server, installed Linux, set up a LAMP stack, and ran the app. If your app got popular, you practiced good load balancing by setting up a second server to ensure the application wouldn't crash from too much traffic.
Times have changed, though, and instead of focusing on single servers, the Internet is built upon arrays of inter-dependent and redundant servers in a system commonly called "the cloud". Thanks to innovations like Linux kernel namespaces and cgroups, the concept of a server could be removed from the constraints of hardware and instead became, essentially, a piece of software. These software-based servers are called containers, and they're a hybrid mix of the Linux OS they're running on plus a hyper-localized runtime environment (the contents of the container).
Container technology can be thought of as three different categories:
-
Builder: a tool or series of tools used to build a container, such as distrobuilder for LXC, or a Dockerfile for Docker.
-
Engine: an application used to run a container. For Docker, this refers to the docker command and the
dockerddaemon. For others, this can refer to thecontainerddaemon and relevant commands (such as podman.) -
Orchestration: technology used to manage many containers, including Kubernetes and OKD.
Containers often deliver both an application and configuration, meaning that a sysadmin doesn't have to spend as much time getting an application in a container to run compared to when an application is installed from a traditional source. Dockerhub and Quay.io are repositories offering images for use by container engines.
The greatest appeal of containers, though, is their ability to "die" gracefully and respawn when load balancing demands it. Whether a container's demise is caused by a crash or because it's simply no longer needed because server traffic is low, containers are "cheap" to start, and they're designed to seamlessly appear and disappear. Because containers are meant to be ephemeral and to spawn new instances as often as required, it's expected that monitoring and managing them is not done by a human in real time, but is instead automated.
Linux containers have facilitated a massive shift in high-availability computing. There are many toolsets out there to help you run services, or even your entire operating system, in containers. The Open Container Initiative (OCI) is an industry standards organization that encourages innovation while avoiding the danger of vendor lock-in. Thanks to the OCI, you have a choice when choosing a container toolchain, including Docker, CRI-O, Podman, LXC, and others.
By design, containers can multiply quickly, whether you're running lots of different services or you're running many instances of a few services. Should you decide to run services in containers, you probably need software designed to host and manage those containers. This is broadly known as container orchestration. While Docker and other container engines like Podman and CRI-O are good utilities for container definitions and images, it's their job to create and run containers, not help you organize and manage them. Projects like Kubernetes and OKD provide container orchestration for Docker, Podman, CRI-O, and more.
When running any of these in production, you may want to invest in support through a downstream project like OpenShift (which is based on OKD.)
The open source components of Docker are gathered in a product called Docker Community Edition, or docker-ce. These include the Docker engine and a set of Terminal commands to help administrators manage all the Docker containers they are using. You can install this toolchain by searching for docker in your distribution's package manager.
One of the great things about open source is that you have choice in what technology you use to accomplish a task. The Docker engine can be useful for lone developers who need a lightweight, clean environment for testing, but without a need for complex orchestration. If Docker is available on your system and everyone around you is familiar with the Docker toolchain, then Docker Community Edition (docker-ce) is a great way to get started with containers.
Dockerhub and Quay.io are repositories offering images for your container engine of choice. If Docker Community Edition is unavailable or is unsupported, then Podman is a wise option. The effort to ensure open standards prevail is ongoing, so the important long-term strategy for your container solution should be to stick with projects that respect and foster open source and open standards. Proprietary extras may seem appealing at first, but as is usually the case, you lose the flexibility of choice once you commit your tools to a product that fails to allow for migration. Containers can be liberating, as long as they're liberated.
This work is licensed under a Creative Commons Attribution-Share Alike 4.0 International License.