So, I’ve been self-hosting for decades, but on physical hardware. I’ve had things like MythTV and an asterisk voip system, but those have been abandoned for years. I’ve got a web server, but it’s serving static content that’s only viewed by bots and attackers.

My mail server, that’s been active for more than two decades is still in active use.

All of this makes me weird in the self-hosted community.

About a month ago, I put in a beefy system for virtualization with the intent to start branching out the self hosting. I primarily considered Proxmox and xcp-ng. I went with xcp-ng, primarily because it seems to have more enterprise features. I’m early enough in my exploration that switching isn’t a problem.

For those of you more advanced in a home-lab hypervisor, what did you go with and why? Right now, I’m pretty agnostic. I’m comfortable with xcp-ng but have no problems switching. I’m particularly interested in opinions that have a particularly negative view of one or the other, so long as you explain why.

  • poVoqA
    link
    fedilink
    English
    arrow-up
    1
    ·
    1 year ago

    If you don’t actually want to allow external untrusted people accessing your server, why go the VM route? That seems like a huge waste of resources and just complicates things compared to using containers (Podman is best IMHO).

    • phase_change@sh.itjust.worksOP
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      I have no problems with untrusted people accessing resources I intend to be public. A VM provides an extra layer of protection in that scenario, as does a container. I’ve been playing with Lemmy containerized in an xcp-ng VM.

      But really, it’s a chance to learn and play with something new.

      • poVoqA
        link
        fedilink
        English
        arrow-up
        1
        ·
        1 year ago

        I mean as in renting out servers (VMs), where untrusted people have full root access.

        • phase_change@sh.itjust.worksOP
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 year ago

          Ah. Yes, I have no plans to do something like that.

          My answer still applies. If there’s a remote code exploit that can be used to gain root, running it in a container just gets you root there. Running it in a VM only gets you root there. Both provide layers to protect the underlying OS.

          • poVoqA
            link
            fedilink
            English
            arrow-up
            3
            ·
            1 year ago

            Indeed, VMs are more secure than containers, but they come had a quite heavy price performance wise and are also harder to maintain. With Podman you can manage containers just like any other systemd service, which is really convenient.