How to Build a Lightweight Private Cloud with Docker and Ansible

Problem Overview

Enterprises often want to run a cloud platform inside their own environment, but replicating a public‑cloud service incurs high cost and low reproducibility. Typical pain points include single‑machine limitations, modest scale (tens of thousands of users), and a complex, heterogeneous technology stack spanning Java, Erlang, databases, and messaging systems.

Lightweight Private‑Cloud Challenges

Insufficient single‑node capacity.

Scale‑out not as high as public clouds (5‑10 万 users).

Very diverse technology stacks make packaging and deployment difficult.

The goal is to simplify deployment while preserving isolation, resource control, and high availability.

Proposed Solution

The solution is built on three principles:

Standard OS provides compute resources – works on both physical machines and virtual cloud hosts.

Docker packages applications and isolates runtime resources – each service, regardless of language (Java, Erlang, C, etc.), is containerized, allowing different JDK versions or library dependencies to coexist without conflict.

Ansible orchestrates distributed deployment – replaces heavyweight Kubernetes for small‑scale private clouds, handling multi‑host configuration, inventory, and playbook execution.

Key Toolchain

Docker – container engine; concepts include images, registries, containers, volumes, networks, and Dockerfiles.

Kubernetes – mentioned for context but omitted from the lightweight stack due to its operational overhead.

Ansible – Python‑based automation framework; uses inventory files, connection plugins, modules, and playbooks to perform batch configuration, deployment, and command execution without agents.

Architecture Diagram

The stack consists of an OS layer, multiple Docker containers, and monitoring agents (MetricBeat, Filebeat, etc.). Supervisor runs inside each host to keep services alive. Docker’s cgroup isolation enforces CPU and memory limits. Playbooks (YAML files) describe how to deploy each service, e.g., DB clusters or object storage (NOS).

Practical Example: RabbitMQ Cluster

A sample repository ( https://github.com/williamleung/lite-deploy-sample) demonstrates how to automate a RabbitMQ cluster using Docker, Ansible, and Vagrant to simulate a multi‑host environment. The process includes node role assignment, cookie synchronization, and user initialization, aiming for a three‑step deployment metaphor: open the fridge, put the elephant in, close the fridge.

Roles and Workflow

Developers build Docker images, write deployment scripts, and distribute artifacts.

Operators provision servers, initialize environments, and execute Ansible playbooks to bring up the cluster.

References

Vagrant box search: https://app.vagrantup.com/boxes/search NetEase Cloud container service: https://c.163.com/hub#/m/home/ Ansible Galaxy: https://galaxy.ansible.com/ Docker mirror:

https://download.daocloud.io/Docker_Mirror/Docker