New "storage" type server for nextcloud

[CropS]

I'm using proxmox for long time, but this is the first time i'm creating the storage system where nextcloud will be installed.

So Hardware is going to be:
CPU: Ryzen 3900
RAM: 128GB
SSD: 2x2TB
HDD: 8x22TB

My plan is to install proxmox and then inside the VM it will be nextcloud installation, but my question is, would ZFS be better for this kind, or stay on typical raid10 (software).
SSD will be the installation of proxmox and caching system. How would you configure ZFS, and type of zraid? Any suggestion is recommended for longer time of survival for this server.

 

[pulipulichen]

If it were my setup, I’d keep things simple and split the roles clearly. I would install Proxmox on SSDs first. If the server has a proper hardware RAID controller with protected write cache, running RAID1 for the OS is perfectly fine. If not, I would just use a small ZFS mirror for the system disks. Keeping the hypervisor on SSD makes the whole system feel much more responsive and avoids having heavy Nextcloud storage workloads interfere with the host itself. In my experience, putting PVE directly on large HDD RAID-Z pools just makes maintenance less flexible later on.

For the data drives, I would configure all large HDDs as a ZFS RAID-Z2 pool. With modern 18TB+ disks, rebuild times are long, and RAID-Z1 simply carries more risk than it used to. RAID-Z2 gives you tolerance for two disk failures, which is much more appropriate for high-capacity drives. On top of that, ZFS provides end-to-end checksumming and protection against silent corruption, which is exactly what you want for a file storage workload like Nextcloud.

As for running Nextcloud itself, I would deploy it inside a Proxmox LXC container rather than a full VM. Containers have lower overhead, use CPU and RAM more efficiently, and allow you to mount ZFS datasets directly without adding another virtual disk layer. Snapshots and backups are also cleaner this way, and migration is generally easier. Overall, this layout keeps things straightforward, scalable, and easier to troubleshoot long term.