My Homelab – Architecture, Design, and Why I Built It

#

This post is the first in a small series documenting my homelab. I use this environment for a mix of self-hosting, media services, automation, monitoring, and general experimentation. It also supports a handful of side projects, club infrastructure, and tools I rely on regularly.

The goal has never been “enterprise perfect.” The goal is:

• Reliable enough to trust
• Flexible enough to experiment
• Simple enough to recover when something breaks

This document is meant to be a living reference for me and a starting point for anyone curious about how it’s put together.

High-Level Architecture

#

At a high level, everything at home centers around a Proxmox cluster. Most workloads run either as VMs, Containers, or Docker containers inside VMs. A Synology NAS handles bulk storage and is also part of the backup chain. Some public-facing services live on a small set of Linode VPS instances.

I try to keep a clear separation between:

• Network edge (firewall / routing)
• Virtualization layer
• Container layer
• Storage
• External VPS

Most inbound access is handled through Caddy as a reverse proxy or Cloudflare Tunnels.

Network Design

#

Hardware

#

• UniFi UDM-Pro (gateway / firewall)
• UniFi Core Switch
• UniFi Office Switch
• UniFi Access Point

Everything network-wise is UniFi, which keeps management simple.

VLANs

#

Most infrastructure lives on the default VLAN. Public services are placed on VLAN 9. Anything risky or experimental goes into VLAN 99. Guests get their own isolated network on VLAN 123.

Security Approach

#

• UDM-Pro handles firewalling
• Cloudflare Tunnel or Caddy reverse proxy used for most public access
• Very minimal direct port forwarding
• Internal services are not exposed unless necessary

Virtualization Layer (Proxmox)

#

All local compute runs in a single Proxmox cluster named Home.

Cluster Nodes

#

Not all nodes are equal, and that’s okay. Heavier workloads tend to land on jc-pve04.

Storage Layout

#

• Local-lvm on each node
• Proxmox Backup Server (PBS)
• Synology NAS – ~20 TB usable, serving SMB shares for file storage, a backup target, and media storage

Container Platform

#

Docker runs inside VMs rather than directly on Proxmox hosts.

• One Docker VM at home
• One Docker VM on Linode

I manage containers using Dockge and Docker Compose.

Core Services

#

Monitoring

#

• Grafana – Dashboards and visualization for metrics collected across the homelab.
• Prometheus – Metrics collection and time-series storage that feeds the Grafana dashboards.
• Zabbix – Infrastructure monitoring and alerting across hosts and services.
• Uptime Kuma – Lightweight uptime/status checks with notifications and public views.
• InfluxDB – Time-series database, mostly used for sensor and IoT data.
• Rybbit – Privacy-friendly web analytics for my sites.

Media

#

• Jellyfin – Self-hosted media server for movies and TV.
• Radarr – Automates movie downloads and library management.
• Sonarr – Automates TV show downloads and library management.
• SABnzbd – Usenet download client feeding Radarr and Sonarr.
• Immich – Self-hosted photo and video backup, like a private Google Photos.
• BirdNET – Identifies bird species from audio picked up around the property.
• ADS-B Tracker – Tracks aircraft overhead using an SDR receiver. Public feed: adsb.joshuacarmack.com.
• Restreamer – Restreams IP camera and RTMP feeds for local viewing and recording.

Automation

#

• Node-RED – Flow-based automation for IoT and integration tasks.
• n8n – Workflow automation for connecting apps and APIs.
• Cronicle – Scheduled job runner for scripts and recurring tasks.

Infrastructure

#

• Caddy – Reverse proxy in front of most self-hosted services, with automatic HTTPS.
• Cloudflared – Cloudflare Tunnel client for exposing services without opening ports.
• EMQX (MQTT Broker) – Message broker for IoT devices and automation.
• Gitea – Self-hosted Git server for personal and club repos.
• Gitea Mirror – Keeps Gitea repos synced with GitHub.
• Home Assistant – Central hub for smart home automation: lights, sensors, and other connected devices.
• FreePBX – Self-hosted VoIP/phone system.

Utility

#

• Paperless-NGX – Document scanning and archiving with OCR search.
• Hoarder – Bookmark and link archiving tool.
• HomeBox – Tracks household/homelab inventory: hardware, spare parts, and other physical assets.
• HomeBox Companion – Companion app for scanning items into HomeBox.
• ITFlow – MSP-style documentation and client management, used for a side project.
• BentoPDF – Web-based PDF toolkit for merging, splitting, and converting files.
• CommaFeed – Self-hosted RSS reader.
• Invoice Ninja – Invoicing and billing for side projects.
• IT-Tools – Collection of handy utilities for developers and sysadmins.
• Networking Toolbox – Networking utilities like subnet calculators and DNS lookups.
• Mixpost – Social media scheduling and posting.
• OnTime – Event and stage timer for livestreams and presentations.

Highlights

#

A few of these get a callout since I rely on them the most day-to-day: Zabbix, Caddy, HomeBox, and Home Assistant (see descriptions above). My ADS-B Tracker is also public if you want to see what’s flying overhead: adsb.joshuacarmack.com.

External Infrastructure (Linode)

#

I use a few small VPS instances mainly for:

• Public Docker workloads
• TacticalRMM
• Offsite monitoring

This keeps important applications offsite with redundancy.

Backups & Disaster Recovery

#

Backup flow:

VMs / Containers
    ↓
Proxmox Backup Server
    ↓
Synology NAS
    ↓
Backblaze B2

This gives me:

• Fast local restores
• Protection from single-disk failure
• Offsite copy for worst-case scenarios

Nothing fancy, just layered.

Operational Practices

#

• Clear naming (jc-pveXX, descriptive VM names)
• If something is annoying to rebuild, it gets backed up

Lessons Learned (So Far)

#

• Clusters don’t have to be identical to be useful
• Backups matter more than perfect architecture
• Containers inside VMs are a good compromise for flexibility
• Documentation saves future-me a lot of time

Roadmap

#

• Better documentation
• Add NetBox or similar for inventory
• More automation around backups and monitoring
• Possibly add another small node in the future

Closing

#

This is a living setup and a living document. I’ll update this as things evolve and plan to do deeper-dive posts on Proxmox, Docker, and monitoring later.