When you run a homelab you might know the situation. You have your lab running but you stumble upon new cool things, you notice some things don’t work the way you want them to or could be optimized. And that is when you start thinking about how to improve your environment or include new requirements.
I learnt this is a never ending iteration – at least for me and I guess for a lot of other people out there, too.
In this series of posts I want to describe my 2022 revamp process and results, so stay tuned for more posts.
I am running a homelab/home environment for many years now and started as a lot of others, too. With an old PC, running as a server. Luckily through my previous job I was able to get my hands on a lot of fine server hardware for reasonable prices so my environment grew and improved.
Here’s what it currently looks like:
These are the things that make me think about the current setup again:
No shared storage
I do not have any real shared storage in the current setup. I can provide storage available in the production host (or even the backup host) via a VM and an NFS or iSCSI share, but I don’t count that as a full shared storage. As soon as I need to take the production host down, I need to Storage vMotion Vms to another host/datastore.
That comes also from the “No shared storage” point – taking my prod host down even disables my possibilities to browse the Internet (DNS & router are on the prod host as VMs). Taking the production host down for maintenace will require manual migration work beforehand (migrating VMs, e.g. to the backup host or failing over to the replicas)
Server hardware is heavy on power usage. These are the current consumption values without any special load:
- Production ESXi: 140-150W
- Backup ESXi: 120W
- Lab ESXi: 180W
To save here I run the backup server only in the backup window at night (3.5h/day) and the lab server only on demand. So the only 24/7 position out of this is my production system at 150W.
I see an issue in the server hardware here. Just the baseboard management controller on the Supermicro board of my production server will consume 14W when the system is otherwise powered off.
For any kind of change in the environment (e.g. updates) my manual efforts to migrate data & services for availability are high. This goes also for my lab server which I only power on at demand. I do have a script to do this automatically, but I have to maintain it to start and stop the VMs in the correct order.
This was even worse in my pre-2020 setup when I had the storage on the lab server provided by a QuantaStor VM on the very same server – shutting down the datastore before the other VMs was a pita.
So that’s enough for today. I have a gorgeous homelab with lots and lots of possibilities, but it’s not perfect and I do have issues which I encounter every day.
I’ll write about my ideas for the 2022 revamp in the next post, so stay tuned.