I have been diving into the world of home labs and am eager to create a setup that not only meets my needs but also allows me to learn and experiment with different technologies.
As a beginner, I am looking for some guidance on the initial steps to take. I am particularly interested in setting up a network environment where I can practice virtualization, containerization, and maybe even dabble in some networking projects. I have some old hardware lying around, including a couple of desktops and a Raspberry Pi, and I am wondering how I can best utilize them.
Additionally, I had love to hear your recommendations for essential software tools that would be beneficial for someone just starting out. Should I focus on specific hypervisors, or is it better to explore multiple platforms?
Hey there @macede1883! It’s great to hear your looking to dive into homelab. For getting started, I think the best overall advice is don’t overdo it. I find it’s really easy to over engineer and get in over your head with too much to fast. At least that was my experience. It turns out that I didn’t actually need 30 vlans like I thought when I got started. I often find simpler is better for getting started and you can always make it more complicated later.
In my homelab, I’m running Unifi equipment for networking, TrueNAS Scale for NAS and storage, and Proxmox as my hypervisor. I think it’s all very friendly, scales well as your homelab changes, and TrueNAS and Proxmox are lightweight enough to run even on older hardware. You’ll also find a lot of love from others on this forum for those two products. XCP-ng is another popular homelab hypervisor. I just can’t speak to it personally.
For managing and maintaining, I started out using what was provided with the products with notifications by email. The ability to get notifications when something needs attention is a great first step. If you end up running debian based distros (including Proxmox), the unattended-upgrades has helped me greatly automate the update security packages on my systems. More recently, I’ve been using ansible to automate other tasks and maintenance which then sends me messages on Discord with the results.
If your interest is virtualization, containerization, and networking projects then you’re probably talking about building out a cluster of small computers where the storage on each node doesn’t have to be huge. To use your old desktops for this effectively their CPUs need to support virtualization technology (VT-x for Intel and AMD-V for AMD, and possibly also VT-d/IOMMU depending on the project).
Since you want to learn about networking, if you don’t already have a switch I’d look for a managed switch rather than an unmanaged one, and with sufficient spare ports to conncet and disconnect things while experimenting. A 1G switch should be fine. You might want to plan to add 10G to the network eventually, but that is mostly a physical layer that doesn’t affect any of the software concepts.
I’d check out the ServeTheHome channel and what they have published over the years about what they call Tiny/Mini/Micro PCs. basically 1L PCs and NUCs. This is probably the cheapest way to build out a cluster.
STH has switch videos as well, but they range from really low end to really high end. I couldn’t quickly find one good reference from them on managed homelab switches. Here’s something from someone else;
As far as which hypervisors or other software, that really depends on if and how immediately or directly the lab is tied to your career. If you have some direct career goals, then that can inform which software you might want to focus on. If the goal is more conceptual, then you should probably be prepared to load different software, tinker with it, and compare and contrast. I guess the two most talked about FOSS options are Proxmox and XCP-NG. You’d have to have some sort of paid license to even experiment with VMWare and HyperV now I think. And of course you could install the necessary KVM, Docker, Kubernetes, etc packages into your own Linux distro, there might be some learning value in that–seeing what all is going on under the hood–rather than working with a preconfigured hypervisor-specific disro.
I find it’s really easy to over engineer and get in over your head with too much to fast. At least that was my experience. It turns out that I didn’t actually need 30 vlans like I thought when I got started. 
I often find simpler is better for getting started and you can always make it more complicated later.