Day 4: Home Lab Infrastructure - Personal Cloud Setup

Today you’ll be exploring ways to build your own personal cloud on your own hardware at home. Therefore escaping the “cloud tax”. As a DevOps your job will also be thinking about things, setting up infrastructure and communicating with management. If you’re serious about enhancing your skills, you will want to consider getting your own hardware.

Owning your own hardware and proving you can manage it is a competitive advantage for any job. It’s a project which will put you ahead of most other developers or sysadmins. It isn’t expensive and will give you the most experience because you can test things and break them as you wish in a safe environment without incurring expensive cloud bills if you forget to destroy a kubernetes cluster which typically would be a minimum of 3 VMs + Master control plane VM.

After you will work with your local infrastructure, you will see that the big cloud providers all provide on top convenience services which makes companies and infra dependent. Therefore, after you become a master DevOps Engineer you will want to simply go back to using virtual machines in the cloud instead and do things yourself;) for better cost savings AND efficiency.

There are open source tools which can help us simulate an AWS cloud locally and we’ll explore these later days.

The refurbished Secondhand Secret

If you have an older laptop, desktop, workstation you can turn that one into your homelab. If you don’t have one you can turn into a “server” then I advise you to buy a secondhand or a refurbished workstation with 64-92 gb ram, 2-3 HDD/SSD drives and 4-12 CPU cores. SO you have enough juice to run a real workload. Similarly if you find cheap Lenovo Thinkcentres you can also get 2-3 of them and simulate your own “datacentre”. Having 1 single juicy workstation is mostly better since you can simulate your datacenter with VM’s anyway:).

Case study in longevity and Return on Investment

IN 2016 I bought a refurbished HP z620 workstation for 2000 euro. 2 CPU 8 core Intel XEON, 92 GB ram to use for virtualization, cyber security, programming, devops, kubernetes etc. Had I bought it NEW with that hardware, I would have spent more than 10.000 euro. Want to know the interesting thing? Even though it was refurbished (second hand) in 2016, it still works perfectly in 2026. I’ve since then only upgraded the SSD drives.

Depending on where you are located take a look at local refurbished shops, in the EU you even get 2 years warranty;). Don’t go spending money on

  1. New hardware - Deprecation is a killer, It’s not worth it
  2. Server racks and blades, etc. They take up a lot of space, before you’re sure you are into infrastructure use workstations since they more power efficient
  3. “Cool gamer aesthetics colored casing” - No one will see it.

NOTE ON RaspberryPI’s: Theoretically you can use raspberrypi’s, however when using docker, containers k8s you will have to manually crosscompile a lot of applications because of the ARM cpu. WHile this is a great way to learn devops, CI/CD it’s also going to frustrate you if you then plan to deploy to cloud as you’d need to keep 2 versions of it.

Price Comparison Homelab vs Cloud AWS

If you’d get a 32GB RAM 8 CPU EC2 VM with AWS, that would amount to $192/ 167 euro /month. If you get a refurbished thinkcentre for example (not affiliated) which is 400 euro it’ll pay for itself after 3 months. Yes, you still pay for electricity or internet at home however you’re learning;)

Get Linux ON it

After you have the hardware, wipe the hdd/ssd clean. It’s now time to install a Linux distribution on your workstation/server. I recommend Debian for the stability and security aspects. It’s rock solid, i’ve been running it on my personal and professional cloud virtual machines for 15+ years. You can go with whichever distro you prefer. I will be using debian and demo commands for it. Since we’ll be using KVM to virtualize other OS’es, the guest operating systems can be anything.

Have your workstation or laptop? Great, jump to day 5 where we’ll be using KVM and LibVirt;)