I use hardware RAID for most of my servers, with a hot spare drive. This is not StartOS specific, but just a general hardware platform default. It has served me well for decades but when you get a hardware drive failure (and I think for spinning HDDs the general Google metric is 8% chance of failure per year), RAID provides me time to recover and replace without losing any uptime.
In the case of software RAID, Proxmox went in hard with ZFS many years ago. It works and gives a similar recovery window. The issue (and this applies to hardware and software RAID) is that you need to be pretty well matched for HDD or SDD capacity. This means that over time, with entropy, hard drives get cheaper and bigger and trying to expect that you can buy replacement drives of the same capacity years later might work but often doesn’t. As a result, I’ve separated my processing from my perpetual data storage.
Although this mitigates having different sized hard drives in an array (ie. Unraid, etc. allow you to have arrays that have differing drive sizes in them), the biggest advantage of doing this is backup recovery. If you have every had to recover a poorly thought through 1TB KVM on a server, you’ll know what I’m talking about. Virtualization should be structured much like the old trick of having your OS on a boot drive, and data on a separate drive.
This is where Proxmox or other virtualization platforms shine. You can set it up to have your OS drive on a local, small RAID setup for resilience and recovery. But move your data drive to a NAS or something that is better suited for data storage. This will likely be more cost effective too. What I do is to have a RAID storage system for my VMs (OS drive) and a NFS share for my data (on a NAS). The crazy thing here is that you could run FreeNAS/TrueNAS for the NAS as a VM in your Proxmox system, and just have its storage be external with a NAS or DAS enclosure, and serve it as a NFS share or something like that to the other VMs (ie. StartOS).
Or just have a separate hardware box for your NAS and use that for everything storage on your network. It works for me. YMMV of course, and there are an almost unlimited number of possibilities on how you structure this. But if your ultimate goal is maximum uptime, having a RAID storage setup on your Proxmox rig, and putting your OS drive on that, and then separating your data drive to something more akin for longer term data storage (ie. a NAS) may be a direction you might want to follow. As the blockchains get bigger, it makes it easier to provision and manage storage this way, without expecting a hypervisor to also be your longer term perpetual data storage solution.
Some thoughts here, but there are many other ways of course. Just keep in mind what happens when the s#$t hits the fan and things fail, and plan your disaster recovery process BEFORE that happens.
