Those old server CPUs and motherboards just draw a ton of power, I’m not sure there’s much more you can do.
It might be best to change hardware if you need lower power draw, a $50 PC off ebay with an i5-7500 should be faster and a only uses about 15W idle.
That N200 is likely on par or faster than dual Opteron 6272 CPUs, since they are so old.
Yup, this is just saying 510mA in a weird way
Essentially a container shares the kernel of the host, so uses less resources to run.
VMs are useful when you need more isolation or a different kernel (or need to add kernel modules).
For most purposes containers are the easy option.
I guess I could have a Container with a bind mount to a dataset that I want to be able to share over NFS or SMB, and handle that from whatever OS I put in the Container, right?
Yep!
But, I could also have a VM do that, and though it wouldn’t be able to share the data with other VMs, it can do it over NFS, can’t it?
Also yes, just a more complex setup with more performance penalty due to using NFS to share data into the VM.
You wouldn’t want to write to the same file at the same time, but otherwise it works fine.
That’s almost worse, Asus also makes a lot of garbage.
That makes sense.
While windows doing a bunch of IO is silly and probably could be more efficient, it sounds like the blame is the SSD controllers crapping out when asked to do exactly what they advertise they are supposed to do (high throughput IO).
Is this 100% an OS issue, or a hardware issue just being made apparent by the OS doing something weird?
Migration should be as simple as importing the existing ZFS stuff into the Proxmox OS. Having backups of important data is critical though.
If I understand correctly, you can specify directories that live on the Proxmox Host, and they appear inside the VM, right?
Inside a Container, yep. VMs can’t do bind mounts, and would need to use NFS to share existing data from the host to inside the VM.
How does this compare to using virtual storage?
Like a VM virtual disk? Those are exclusive to each VM and can’t be shared, so if you want multiple VMs to access the same data then NFS would be needed.
But containers with bind mounts don’t have that limitation and multiple containers can access the same data (such as media).
Also, how can I replicate the ZFS pools to an external machine?
ZFS replication would do that.
Proxmox supports ZFS natively with management in the WebUI. So you could get rid of TrueNAS entirely and not need to deal with HBA pass-through or anything.
You also wouldn’t need NFS or have to deal with shares, as the data is available directly to Proxmox Containers via bind mounts.
Newer CPUs can also just be better optimized and have more faster cache and that sort of thing, so might be faster at running a process even if they’re the same on paper.