Why We Self-Host Everything (And You Should Consider It Too)

Every month, we pay ~£40 for a dedicated server that runs our entire production infrastructure. The equivalent on AWS, Vercel, and managed databases would cost £500-1,000+. Here's why we made this choice.

What We Run

• PledgeNow — Next.js app + Postgres database
• QuikCue website — Static HTML + nginx
• 2 Ghost blogs — with MySQL
• JustVitamin — Python Flask + Postgres
• Gitea — Self-hosted Git
• Communication stack — Chatwoot, n8n, WAHA (WhatsApp)
• Monitoring — Uptime checks, log aggregation

Total: 15+ containers across 6+ applications, serving real users in production.

The Cost Comparison

That's a 10-25x cost reduction. Over a year, we save £4,000-12,000. For a startup, that's runway.

The Advantages Beyond Cost

1. Speed of Debugging

When something breaks at 2am before a charity's event, we SSH in and fix it. No support tickets. No waiting for AWS to respond. No "escalating to the engineering team." We see the logs, fix the issue, restart the container. Done in minutes.

2. Full Control

We own every layer of the stack. No vendor surprises:

• No sudden pricing changes (looking at you, Heroku)
• No platform deprecations
• No rate limits or usage caps
• No vendor lock-in

3. Learning

Running your own infrastructure teaches you things that managed platforms hide. Networking, DNS, SSL, container orchestration, database management — all skills that make you a better engineer.

4. Privacy

Donor data stays on our server. Not on AWS (which serves the US government). Not on Vercel (which processes data through Cloudflare). Our data, our responsibility, our control.

When Self-Hosting Doesn't Make Sense

• You don't have anyone who can SSH into a server — self-hosting requires operational knowledge
• You need massive scale — if you're handling millions of requests, managed cloud auto-scaling is hard to replicate
• Compliance requires managed services — some regulations mandate specific hosting environments
• Your time is worth more than the cost difference — if you're a funded startup with engineers billing at £150/hr, managed services save time

Our Stack

• Server: Dedicated (Hetzner, 64GB RAM, 8-core AMD)
• Containers: Incus (lightweight VMs) + Docker (application containers)
• Reverse proxy: Traefik (automatic SSL, routing, load balancing)
• Deployment: Dokploy (self-hosted PaaS)
• Git: Gitea (self-hosted)
• Monitoring: Built-in Dokploy + external uptime checks
• Backups: Automated volume snapshots + off-site database dumps