so right now my homelab consists of a single Dell R620, and a whitebox NAS. My employer “threw out” several other r620s all the same spec (150GB Ram, 2X 6Core Xeon) plus some other servers. their in my rack but turned off. In the spring my employer is going to be throwing out many micro HP EliteDesk pcs & one more R620.

this would mean I would possibly have 4 R620s, A whitebox nas, R620, & a R510. Only one 620 and my NAS run at a time 24/7 I find all the other hardware excessive… heat power and noise are the limiting factors…

Right now I run plex, homeassistant, and a dozen or so other apps on the single r620 but the expansion limits are a limiting factor, I want A GPU for some Machine learning projects.

I am at a crossroads. should I decommission my single r620 for a handful micro PCs, upgrade them to 32GB of ram each and run a lower power cluster? I’d sell the servers and build a whitebox GPU server to go with my whitebox NAS…

​

Thoughts? would it be stupid to sell my 3 or possibly 4 R620s for some lower spec hardware?

  • Plane_Resolution7133@alien.topBEnglish
    1·
    10 months ago

    I would keep one ‘proper’ server, 3-5 SFF machines for cluster labs and such.

    One of the SFF could be a decent 24/7 production machine with light Proxmox duties or something similar.

    • alphagatorsoup@alien.topOPBEnglish
      1·
      10 months ago

      Greatly appreciate the input! I’m still trying to decide! I want to limit my power use and noise but improve expandability and potential for new projects. The lack of GPU has been super annoying