Building a segmented cybersecurity home lab with Proxmox
I built a Proxmox home lab to explore how network boundaries, Windows security policies and monitoring work together. OPNsense separates the lab from my home network, Active Directory manages a Windows 11 workstation, and an Ubuntu Docker host gives me visibility into the infrastructure.
Separating the lab from my home network
I configured two virtual bridges in Proxmox. The first connects to my home and management network, where I manage the hypervisor and run the Ubuntu monitoring server. The second connects the Windows and Kali machines to a dedicated lab network. OPNsense acts as the lab gateway, provides DHCP and controls traffic crossing that boundary. I used Kali to verify connectivity to the gateway, domain controller, Windows client and internet. On the Windows client, I kept the firewall enabled and allowed ping through a single inbound ICMPv4 echo rule scoped to the Domain profile. That gave me a way to test connectivity while preserving the workstation’s firewall baseline.
Managing Windows security with Active Directory
I set up a Windows Server 2022 domain controller with Active Directory Domain Services and DNS, then joined a Windows 11 Enterprise workstation to the domain. I created separate organisational units for users, workstations, servers, groups and service accounts so I could apply policies to the right systems. I split workstation policy into two Group Policy Objects. The first defines the security baseline, including a 15-minute inactivity lock, a disabled Guest account, UAC prompts on the secure desktop, Microsoft Defender and Windows Firewall. The second handles auditing. It records successful and failed sign-ins, account lockouts, security group changes and process creation. With command-line capture enabled, Windows Event ID 4688 gives me a concrete record to examine when investigating what ran on the workstation.
Monitoring the lab with Docker, Prometheus and Grafana
The Ubuntu server hosts my Docker monitoring stack. Prometheus collects metrics, and Grafana brings them into a dashboard covering three layers: the Proxmox VE exporter reports on the virtualisation environment, Node Exporter provides Ubuntu system metrics, and cAdvisor shows resource use by container. Uptime Kuma checks whether monitored services respond. Those layers help me narrow down a problem. If a service responds slowly, I can compare its container’s CPU and memory history with the load on Ubuntu and the physical host over the same period. If the question is who signed in or what executed, I have Windows audit logs to examine.
What I took away
This project changed how I verify a security control. A successful ping tells me one specific kind of traffic reached the workstation. A Group Policy setting tells me what I configured. An event in the Windows Security log shows what the workstation actually recorded. Having all three perspectives in one lab gives me a practical way to test an assumption and explain the evidence behind it.