A Datacenter at Home · Cason Miles

Problem

You do not learn enterprise infrastructure from a course. You learn it by being the person responsible when the hypervisor misbehaves, the backup silently fails, or the firewall rule you wrote at midnight locks you out. So I built the environment that forces that learning: a homelab run with production discipline, on the same classes of tools enterprises run.

FIG 01 · SCHEMATIC · LAB TOPOLOGY

Multi-monitor desk setup with one screen actively running a live Netdata monitoring dashboard alongside the daily workstation displays, keyboard, and speakers — FIG 02 · OPERATIONS DESK

Every dashboard on this page is a live pane I actually look at, not a screenshot staged for the portfolio.

Approach

The backbone is Dell PowerEdge server hardware with iDRAC out-of-band management, running a VMware vSphere environment (vCenter over ESXi) that hosts a fleet of Debian and Ubuntu virtual machines.

A Cisco Catalyst 3750X sits at the core carrying five VLANs: services, users, guest, VPN and an isolated management network. It is deliberately pure layer 2, so every inter-VLAN flow hairpins back to the edge and gets inspected rather than silently switched.

That edge was OPNsense, running as a virtual machine on the very hypervisor it protected, doing firewall, DHCP, recursive DNS and reverse proxy for the whole estate.

Veeam Backup & Replication protects it all, because infrastructure without tested backups is a hobby, not a platform.

Two Dell PowerEdge R730 servers stacked on a Dell EqualLogic PS6500 SAN chassis, drive bay labels and activity LEDs visible — FIG 03 · RACK ELEVATION

Netdata overview dashboard on a wall-mounted monitor showing live ESXi host and per-VM CPU, memory, and network telemetry across the vSphere fleet — FIG 04 · LIVE TELEMETRY

vSphere Client performance overview for the ESXi host showing real-time CPU, memory, memory rate, and disk charts across the VM inventory — FIG 05 · VCENTER PERFORMANCE

Portainer home screen listing three connected Docker environments, each showing live stack, container, and image counts — FIG 06 · CONTAINER FLEET · PRIOR GENERATION

Veeam Backup and Replication inventory listing protected virtual machines with guest OS and last-backup timestamps — FIG 07 · BACKUP INVENTORY

Portainer managed the container stacks across the fleet here, and the Veeam inventory shows backup jobs completing on schedule. Portainer has since been retired: it shows what is running but has no concept of what should be running, so I replaced it with a GitOps control plane that deploys every stack from a git repository. That migration is its own case study, Retiring Portainer for Komodo.

OPNsense firewall dashboard showing system health, gateway status, interface traffic split, firewall rule hits, and running services — FIG 08 · EDGE FIREWALL · PRIOR GENERATION

OPNsense held the edge for years, and that dashboard is the one I actually watched. It has since been retired too, for the same class of reason Portainer was: it was a virtual machine acting as the router for the network its own hypervisor depended on, which is a circular dependency you only discover on the day something goes wrong. It was replaced by a physical UniFi Cloud Gateway Fiber, and that cutover is its own case study.

On top of that platform run more than twenty self-hosted services I actually depend on daily: Nextcloud for files, Vaultwarden for secrets, Jellyfin for media, Frigate for camera NVR duty, Home Assistant for automation, n8n for workflow automation, monitoring dashboards for visibility, and local GPU-backed LLM inference for private AI workloads.

Nextcloud dashboard showing recommended files and a personal to-do list inside the self-hosted file and task workspace — FIG 09 · FILES + TASKS

Home Assistant dashboard with room automations, lighting scenes, WLED strip controls, and the full Hue light inventory — FIG 10 · HOME CONTROL

Result

The lab is my proving ground and my production environment at once. Every skill on this site traces back to it: virtualization, network security, storage, backup strategy, monitoring, and the operational habit of documenting and automating instead of clicking through problems twice. When I say I can run infrastructure, this is the standing evidence, powered on right now, serving my daily life.