How to Build a Hyper‑Converged Container Platform on Nutanix: A Step‑by‑Step Guide

1. Hyper‑Converged Infrastructure Overview

Hyper‑Converged Infrastructure (HCI) integrates compute, networking, storage, and virtualization on a single x86 platform, adding features such as cache acceleration, tiered storage, deduplication, compression, and snapshots. Multiple nodes can be network‑aggregated for seamless scale‑out, forming a unified resource pool.

Traditional four‑layer architectures (Ethernet, compute, storage network, storage) face challenges in large‑scale cloud environments, including I/O bottlenecks, management complexity, costly expansion, and centralized data storage.

HCI adopts a Web‑Scale design, merging compute, storage network, and storage layers into a single hardware tier, simplifying data‑center flexibility and reducing complexity.

2. Nutanix Hardware Architecture

The Nutanix 3000 series is a 2U x86 rack server. Each node can include:

1‑4 compute nodes

1‑2 SSDs (400‑800 GB, up to 6.4 TB)

1‑4 HDDs (1 TB each, up to 16 TB)

1‑2 Intel CPUs (up to 64 cores)

64‑512 GB RAM (max 2 TB)

64 GB SATADOM with pre‑installed hypervisor (AHV, ESXi, or Hyper‑V)

Two 10 GbE NICs for storage, two 1 GbE NICs for applications

VGA, IPMI, USB ports, dual power modules, fans

This standard x86 server packs compute and storage, enabling rapid scaling by simply adding a powered, network‑connected unit.

3. Nutanix Software Architecture

Nutanix software consists of two main products:

Prism : a distributed resource management console similar to VMware vCenter.

Acropolis : a multi‑resource manager comparable to vSphere + vSAN.

Key components:

DFS (Distributed Storage Fabric) : provides pooled storage.

AMF (App Mobility Fabric) : enables workload migration between hypervisors.

AHV (Acropolis Hypervisor) : KVM‑based hypervisor.

These components embody the “software‑defined everything” principle, forming the foundation of a Software‑Defined Data Center (SDDC).

4. Container Support on Nutanix

Nutanix AOS 4.7 introduces two Docker‑related services:

ABS (Acropolis Block Services) : exposes distributed storage as iSCSI block devices for VMs or containers, enabling persistent data.

ACS (Acropolis Container Services) : integrates Docker with Nutanix, providing volume plugins for container storage.

ABS stores container data as external volumes, while ACS supplies a Docker‑compatible volume driver (nutanix) for seamless persistence.

5. Container Cloud Management Platform (DCE)

The DaoCloud Enterprise (DCE) platform manages container clusters using Docker Swarm as the orchestrator. It provides application orchestration, image registry, load balancing, logging, monitoring, and network/storage control.

Topology uses a Manager‑Worker model:

DCE Controller : control plane handling user requests.

DCE Agent : worker nodes running containers.

6. Deployment Practice

6.1 Host Generation

Docker hosts are created with the Nutanix Docker Machine Driver, ensuring consistent hardware and software configuration.

Example command (illustrated in the image):

# docker-machine create --driver nutanix --nutanix-url https://<em>controller</em>:9440 --nutanix-username admin --nutanix-password **** my-nutanix-host

6.2 Block Device Allocation

Each Docker host receives an external block device from ABS for Docker Engine storage, configured in direct‑lvm mode to avoid the default loop‑lvm setup.

6.3 DCE Cloud Management Deployment

Use DCE’s Toolbox to install the control plane and join worker nodes:

# bash -c "$(docker run --rm daocloud.io/daocloud/dce install)"   # install controller
# bash -c "$(docker run --rm daocloud.io/daocloud/dce install --force-pull --replica --replica-controller MASTER_CONTROLLER_IP)"   # install replica controller
# bash -c "$(docker run --rm daocloud.io/daocloud/dce join --force-pull MASTER_CONTROLLER_IP)"   # join worker node

Replace MASTER_CONTROLLER_IP with the actual IP address of the primary controller.

6.4 Application Publishing

Deploy a WordPress application via DCE’s app store. Modify the generated docker‑compose.yml to use the Nutanix volume driver:

version: '3'
services:
  wordpress:
    image: wordpress:latest
    volumes:
      - wpdata:/var/www/html
    ports:
      - "80:80"
  mysql:
    image: mysql:5.7
    volumes:
      - wpdata:/var/lib/mysql
volumes:
  wpdata:
    driver: nutanix
    driver_opts:
      volume_driver: nutanix

6.5 Status Check

Verify container status and external volume creation:

WordPress and MySQL containers are distributed across two agent nodes and running.

External volume wpdb is successfully created in the storage pool, providing data persistence.

7. Conclusion

The guide demonstrates how Nutanix hyper‑converged infrastructure can serve as a robust foundation for container workloads, combining hardware simplicity with software‑defined storage and networking. By following the steps above, practitioners can deploy a production‑grade, highly available container management platform with persistent storage, illustrating the practical benefits of a “dual‑engine” (hyper‑converged + container) model.