I am really interested in seeing how far we can take these cheap, low-power devices. With a typical 11W idle, they really appeal to me as a foundation to build upon.
Wanted to ask you all whether there are any technical reasons why we cant have a NAS Killer using the intel HP 290 or ProDesk 400 G4?
Some issues I see with (an attempt at) building a NAS Killer using the intel HP 290 or ProDesk 400 G4:
space? because these come in small cases, there’s barely space for drives for storage - something we need a LOT of in a NAS.
The hack I propose to address this, is to use:
i. expansion cards like this: https://www.amazon.com/dp/B07SZDK6CZ
ii. mount the additional drives externally in a DIY enclosure. This is not the most compact nor aesthetically, pleasing, but unless I’m missing something, completely functional
limited stock RAM? - the HP 290 comes with 4GB stock while the ProDesk 400 G4 comes with 8GB stock. Ideally, we would (possibly?) like to have 2x8GB = 16GB instead to handle the file buffers. Thankfully, both support that expansion.
PSU might be underpowered? - these come with a 290W peak stock SMPS. I am likely worrying unnecessarily here, because even if I connect 6 external 2TB HDDs, each drawing 5W, that’s 30W, plus a 20W idle (likely far less)/40W peak for the system itself, for a total less than 100W, which is 1/3rd of the PSU’s ability.
My use cases:
Build a “NAS” that’s a JBOD using Proxmox, SnapRAID (parity), MergerFS (pool), openmediavault/OMV. I am thinking of using EXT4 FS as the underlying FS for now but if any of you suggest something else, like ZFS, let me know
While either of the intel HP 290 or ProDesk 400 G4 are QuickSync enabled and can do transcoding, I really intend to stream documents/text files/PDF and audio (music, audiobooks) out of the drives - the whole point is to get these drives on the network so 3 computers and 5 phones on the network can access them. Hence, these boxes really will just be serving content off the drives
If there’s some capacity available, I would like to run a OpenWRT/r00ter container on this
Power is pretty expensive (50c/kWh) where I live: every additional watt costs me $5/year. A 10-20W idle, that’s realistic on both these cheap, low power devices, would cost me ~$80/year. Both the low initial amount (~$100) and the ongoing utility expenses (~$80/year) are what attracts me to these machines.
More powerful machines will probably double this budget?
So what do you all think?
Doable or insane/poorly thought out/unrealistic?