I have around 30TB of data scattared around between cloud services, external hard drives, a small plex server, etc.

I would like start a storage server to store everything, with a bit of computing power for other task (plex as mentioned).

I’m looking on ebay at the Dell Edgeserver t330 or similar, starting with 3x 18TB drives, and since it has 8 bay, I’ll have plenty of space for future upgrades. It seems have also a decent cpu, so it should handle plex and some other tasks without issues.

My concerns is about noise; it is going to be put under a desk in my lab and not inside a rack, I saw some videos on youtube and it doesnt seem too noisy, but hard to tell from a video. And for power usage, it will run mostly at idle I think, so I dont think it will drink too much.

Do you think it make sense? Any suggestion?

As said, the only hard requirements are quite a bit disk spaces for future expansion (6x-8x 3.5 drive support) a decent cpu, and doesnt have to be too noisy.

I’ve also evaluated buying a parts a built it from scratch, using like the new intel n100 with is very powerful with low power usage, a big case and so on for a similar price.

  • puppynosee@lemmy.world
    link
    fedilink
    English
    arrow-up
    5
    ·
    1 year ago

    The current gen processors are hilariously efficient if you build the rest the system properly. I will warn you that the low watt rabbit hole gets crazy. Simply changing a component from one slot to another can make a large difference on idle power draw. https://mattgadient.com/7-watts-idle-on-intel-12th-13th-gen-the-foundation-for-building-a-low-power-server-nas/ is a nice write up on a low power build that lists some of the things to consider when it comes to power usage and CPU idle states. At this point, I would stay away from server hardware unless you have a specific reason to run it. It may be cheap to buy at first, but you end up paying a lot more in the long run on power usage. I would also recommended you look at unraid. It has a somewhat proprietary raid format that does the parity calculations on the file level and not the block level. The result is that unraid is able to keep drives spun down and off much more aggressively than a traditional raid setup. It only needs to spin up the drive that has the file and not the whole disk pack.

    • 7Sea_Sailor@lemmy.dbzer0.com
      link
      fedilink
      arrow-up
      1
      ·
      1 year ago

      Do you by chance have an opinion on Snapraid? The fact that I can easily add more drives to my pool down the road is a BIG win for me, and I can deal with the lost performance from not using “classic” Raid - but I’m not yet super sure if SR is stable and all.

      • puppynosee@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        1 year ago

        I do not. I think unraid supports adding arbitrary numbers and sizes of drives, but the largest drive is always the parity drive. Under the hood I think unraid is just mergefs or something. If you really want to roll your own Linux thing and have the same functionally you can, but the management is a lot harder. I am in the arm chair engineering stages of the next iteration of storage in my homelab so I will definitely check out snapraid.

        • 7Sea_Sailor@lemmy.dbzer0.com
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 year ago

          Thank you for your insight. I skimmed the Unraid documentation and found that the Unraid Arrays apparently support the adding of drives to an existing pool whenever you please, which is great. To be honest, I’m just trying to find reasons to not have to buy Unraid, but it seems harder and harder to justify. I was planning to go either SnapRaid or ZFS with OMV because it seems simple and free, and OMV has some nice features… but now I’m not so sure anymore.