Introduction

Building your own virtualised network forces you to learn the hard stuff—routing, high availability kubernetes cluster, monitoring—without waiting for budget sign-off. I spent 358 hours on a diploma project to prove it. Here’s the what-and-why before we dive into the how.

1. The pain points with traditional monoliths

Tight coupling, ugly scaling, vendor lock-in.
Real-world risk: one box dies, everything dies. (Ask me about that server that fell over at 2 AM.)

2. Design restrictions I set myself

Restrictions	How I met it
Budget ≈ €0	100 % FOSS stack (pfSense, K3S, Zabbix, Ubuntu Landscape).
Must survive node loss	3-node control plane + HAProxy VRRP fail-over.
Must be observable	Zabbix

3. Planning & risk snapshot

358 h total effort split across research, build, test and writing.
Biggest risk: rookie mistakes + lab hardware dying. Mitigated via snapshots and lots of caffeine.

4. High-level architecture

pfSense firewall/IDS on the edge.
VMware layer hosting ten VMs (firewall, repo, GUI, 2×Load Balancer, 3×control-plane, 2×worker node).
K3S cluster inside a flat LAN, fronted by HAProxy virtual IP.
Zabbix server in DMZ.

5. What’s next

Part 2 drops next week—full VM build script and pfSense rule-set.

Call to action: Have you built something similar? Drop a comment or DM me on LinkedIn.

Scalable Virtual Network Lab