You already have a working solution
Well...but that's the thing, it's not
really working for me, hence why I am investigating an alternative system/proposal that would be optimised for cost, raw/usable capacity, latency, bandwidth, physical space, power, and noise.
and if the "politics" of the money prevent you from putting together a sane configuration you're just throwing money away for no good reason.
But isn't that where the "magic" happens where people come up with innovative, cost effective solutions with minimal compromises?
I mean, people have said that my 3-node Proxmox HA/EC (2,1) cluster isn't recommended, but given a $450 USD budget, and an idle power budget of just 21 W, no one else has been able to put together a better proposal than what I have currently deployed, where I
still get 1152 GB of raw capacity left and 768 GB of usable capacity (as a result of it being an EC (2,1) configuration.
If there are other proposals where it meets the same financial budget, power budget, and maximises raw/usable capacity, I'd certainly evaluate it as an alternative, but no one has put together anything that can rival what I've deployed yet.
Thus, given those constraints and the maximisation optimisation problem, it might not be a recommended solution, but no one else has come up with a better proposal to replace it yet neither, so it can't be
soooo bad.
(And I also get the CPU performance of three N95 processors, which isn't much, but it's also not nothing neither.)
With the Supermicro 4U, 4-node system, it's got dual Xeon E5-2660 v3 processors (10-cores/20-threads per CPU, 20-cores/40-threads per node, 80-cores/160-threads total), so that's going to be no slouch when it comes to total, raw performance in a package that currently costs $700 USD.
I don't know what a sane solution would look like where you have four nodes, 24 3.5" HDD bays, and a total of 80-cores/160-threads of compute horsepower as well for $700.
