We (my team) is exploring: would OpenNebula KVM/LXD/Firecracker provide a feasible, “drop in” replacement for Proxmox VE?
The key point: we do not, currently have a large, “enterprise” configuration. For us, a single, small (64GB, 6 core) ProxmoxVE (PVE) hardware-server “just works” for our production usage; PVE is a relatively-simple install, the administration is not always that straightforward (OpenNebula seems to have far better documentation), but after we dig around a while, we can usually figure it out and/or get support from PVE’s forums and/or corporate support.
Our initial OpenNebula KVM install ( https://docs.opennebula.io/minione/ ) did not yield a “we can get this work work out-of-the-box in a few mins or hours” experience. First problem: it required “firewall hole punching (within the hypervisor OS) of OpenNebula’s firewall” to make KVM’s visible. (No firewall management is required in PVE until you want it; again, when you being a PVE kvm online it “just works” without much if any additional configuration, for better or worse.) This gives us the feeling that it’s going to be a longer, harder effort to make OpenNebula work for us.
Long term, as our project scales, we suspect we may want to move beyond PVE to a OpenStack-ish, OpenNebula-ish “enterprise” environment. If we could do it now and save a “migration” in the future, that’d be nice. For now: it seems like a lot more work for us to spinup.
But before we discard OpenNebula for our near-term purposes, we ask: would anyone care to offer some perspective as to why we should/would consider OpenNebula a reasonable PVE replacement in our (above) context/evironment?