Building a Real Server Lab to Pass a Real Exam

The problem with studying from slides

I've taken enough IT courses to know how the pattern goes: the instructor reads bullet points, you memorize definitions, and two weeks after the exam you can't remember what ECC RAM actually does. The Server+ curriculum is different — it names real products, real RAID controllers, real error codes. You can't study it properly without touching the hardware.

So when this capstone came up, I didn't want to spin up a few VirtualBox VMs and call it a server lab. I wanted the thing that the exam was actually describing.

The hardware: HP ProLiant ML350p Gen8

The ML350p Gen8 is a tower server that HP shipped between 2012 and 2016. It's out of support, which means you can find them on eBay for a fraction of what they cost new — mine came with dual Xeon E5-2620 CPUs and 48 GB of ECC RDIMM already installed.

It ships with the HP Smart Array P420i RAID controller built into the motherboard. That's the controller the Server+ objectives reference when they talk about battery-backed write cache, ORCA (F8 at POST), and TLER-aware drives. Configuring RAID 5 across the four 1 TB SAS drives took about ten minutes — and it taught me more about parity calculations than any slide deck.

Proxmox VE as the hypervisor

The course objectives cover both VMware ESXi and Hyper-V, but the capstone allowed any Type 1 hypervisor. I chose Proxmox VE for two reasons: it's free, and it exposes everything — the Linux bridges, the KVM/QEMU config files, the ZFS/LVM storage pool structure.

Proxmox classifies each network bridge as a Linux bridge interface (vmbr0, vmbr1, vmbr2). That maps directly to the VLAN and network segmentation concepts in the exam. When the Server+ objectives say "configure virtual switches to segment network traffic," I was literally editing /etc/network/interfaces to create those bridges.

Three bridges, three security zones

The network topology ended up as the most educational part of the whole project. Three bridges, three IP ranges, and a pfSense VM in the middle:

• vmbr0 (10.10.10.0/16) — management network, connects to the physical campus uplink and hosts the Proxmox web UI
• vmbr1 (172.16.0.0/24) — DMZ, where the pfSense WAN interface and Jump Box live
• vmbr2 (192.168.0.0/24) — private LAN, isolated from the outside world; only reachable through pfSense rules

The Jump Box concept — a hardened bastion host that's the single entry point to the private network — is listed in the Security domain as a network hardening technique. Building one in the lab made it concrete: SSH to port 2222 on the DMZ IP, land on the Jump Box, then use it to reach 192.168.0.x addresses on the private LAN. No direct lateral movement from the outside.

Windows Server 2022 and AD DS

The Administration domain dedicates significant weight to Windows Server roles. I deployed Windows Server 2022 Datacenter on vmbr2 at 192.168.0.10 (static IP — the exam is explicit that AD DS requires a static IP before you promote the DC).

The DNS integration that happens automatically during DCPromo is one of those things that's easy to memorize but easy to misunderstand. Running it on real hardware, watching the SRV records appear in DNS Manager after promotion, makes the dependency obvious: Active Directory needs DNS to function, and the built-in DNS server creates the _ldap._tcp and _kerberos._tcp records that clients use to find the DC.

iLO 4: managing a server without a monitor

Out-of-band management tools (iLO, iDRAC, IPMI) show up in both the Hardware and Troubleshooting domains. The HP iLO 4 built into the ML350p lets you attach to the server over the network — virtual console, virtual media, power controls — even when the server is off.

I used iLO to mount the Proxmox ISO and install the hypervisor remotely. That's exactly the use case the exam describes: "manage a server when the OS is offline." No KVM switch, no physical access to the server room required.

What studying from a real lab gives you

The Server+ exam asks scenario-based questions — "a RAID array has one failed drive; what is the current state of the data?" Those questions are easy when you've watched a real RAID rebuild on real hardware. They're harder when you've only read the definition.

Every time I ran into something unexpected in the lab — a DHCP lease not assigning, a VM not reaching the internet, an SSH connection refusing — I was running through the troubleshooting methodology the exam tests: identify, theory, test, plan, implement, verify, document. Not because I was following a checklist, but because that's the only logical way to solve a real problem.

That's the point of the capstone. The exam is a proxy for knowing how to run a real server environment. If you can build the lab, you already know most of what the exam is testing.