Hello all!
I am currently planning out my first homelab and I have been having an immense amount of trouble finding the right configuration for my first home server. I have been bouncing back and forth between platforms and I just feel nervous about making the wrong choices.
My needs really aren’t so crazy, the main purpose of this server will be NAS, and so I will likely put a good bit of the budget into storage. Though, I would also love to run some docker containers here and there for things like Jellyfin, Pi-hole, and Home Assistant. I also would be running some other random Linux VMs, but nothing too critical.
The only three things I really care about otherwise are decent hardware transcoding, power efficiency, and support for ECC memory.
I am considering picking up an older Kaby Lake i3 7100 or maybe a newer i3 10100 and going from there, as the base system would be rather inexpensive this way. But part of me also wonders if I should step up to something with 6 cores. Or maybe there is another option all together that is better?
You don’t need ECC memory for any of this.
ECC memory is rarely beneficial to the home server.
But I digress, moving on. Forget the old hardware recommendations. Especially anything Ivy/Sandy/Haswell/Broadwell. If it’s not Skylake or newer, don’t bother.
That said, I wouldn’t bother at all. No sense in roping yourself in to a dead end platform with poor performance and high power consumption, while simultaneously limiting your expansion possibilities.
$500.
All brand new parts, all very expandable. Plenty of performance too. Compared to the 7100 you were considering it has over 3 times more compute power, while having an obscenely good iGPU for transcoding, should you need it.
And it gives you an upgrade path. Need more compute or more than (8) 4K transcodes? Slap a 13500 in it and go to town. More drives? The R5 gives you 10 drive bays. Etc etc etc.
You couldn’t pay me to go down the used hardware route at this point, not when you can build an entire brand new machine that will significantly outperform them while using less power for $500.