Robots Atlas>ROBOTS ATLAS
Infrastructure

Docker

2013ActivePublished: 19 November 2026Updated: 19 November 2026Published
Key innovation
Popularized lightweight OS-level containerization: packaging an application with its dependencies into a portable image that runs identically on any host, without the overhead of a full virtual machine.
Category
Infrastructure
Abstraction level
System
Operation level
DeploymentServingAgent runtime
Use cases
Reproducible development environments and CI/CDContainerizing and deploying applications (microservices)Isolated sandbox for running AI-agent-generated codeServing and packaging AI/ML modelsMLOps pipelines and reproducible experimentsPortability between cloud and on-premise

How it works

1) An image is built from a Dockerfile — a layered, read-only template (each instruction is a separate, cached layer). 2) The image is pushed to a registry (docker push) and pulled on the host (docker pull). 3) docker run creates a container from the image — a host process isolated by namespaces (its own PID, network, and mount view) and constrained by cgroups (CPU, memory). 4) A union filesystem (OverlayFS) stacks the image layers with the container's writable layer. 5) The dockerd daemon manages the lifecycle, delegating execution to containerd and runc. Containers share the host kernel, so they start in seconds and are lighter than virtual machines.

Problem solved

Eliminates the "works on my machine" problem: discrepancies between development, test, and production environments. It provides portability and reproducibility without the overhead of full virtualization, plus process isolation on a single host.

Components

Docker Engine (dockerd)Runtime / server

The daemon managing images, containers, networks, and volumes; serves the REST API.

Image (OCI image)Distribution artifact

A layered, read-only template with the application and dependencies, built from a Dockerfile.

ContainerUnit of execution

A running instance of an image isolated by namespaces and cgroups.

DockerfileBuild definition

A declarative file defining image build steps; each instruction creates a layer.

Registry (Docker Hub)Image distribution

Image store; the public Docker Hub by default, with private registries possible.

containerd + runcLow-level runtime

Low-level container runtime; runc is the reference implementation of the OCI standard.

Implementation

Implementation pitfalls
Data persistence in containersMedium

Changes not written to a volume/bind mount are lost when the container is removed.

Fix:Use named volumes or bind mounts for persistent data.
Shared host kernelHigh

Containers share the kernel, so isolation is weaker than in VMs — relevant when running untrusted code.

Fix:For untrusted code use extra layers (gVisor, Kata Containers, micro-VMs) and privilege restrictions.
Image size bloatLow

Suboptimal layers and lack of multi-stage builds inflate image size and pull time.

Fix:Use multi-stage builds, minimal base images, and .dockerignore.

Evolution

Original paper · 2013 · dotCloud / PyCon US 2013 · Solomon Hykes
Docker: public announcement and open-source release (dotCloud, PyCon 2013)
Solomon Hykes
2013
Launch and open-source release
Inflection point

Solomon Hykes presents Docker at PyCon; the dotCloud project is open-sourced (March 2013) and dotCloud is renamed Docker, Inc.

2014
libcontainer replaces LXC

Version 0.9 replaces the default LXC backend with its own Go component, libcontainer.

2015
Open Container Initiative and runc donation
Inflection point

Docker and others announce the Open Container Initiative; Docker donates runc as the reference runtime implementation.

2016
Engine rebuilt on containerd + runc

From version 1.11, Docker Engine is built on containerd and runc.

Execution paradigm

Primary mode
Conditional

Not applicable to ML models — Docker is a containerization runtime, not a neural-network architecture.

Hardware requirements

Primary

Containerization is independent of the accelerator; isolation relies on the Linux kernel, not specific hardware.