Any shop large enough that this is such a massive undertaking is large enough that the people who care about this aren’t the people making financial decisions.
The good news is this is horrendous for finance as well so unless you cut a deal for your licensing costs because you’re a titan, you’ll be switching.
Actually, KVM and Ansible are a big part of the industry. KVM as a hypervisor is widely acknowledged in enterprise environments. Meanwhile, Ansible provides great automation for system management tasks. Their combined use for virtualization management and provisioning is not just common but a best practice, backed by a plethora of successful deployments across various industries.
The suggestion to just use KVM and ansible is rather tone def.
It’s also just dumb.
Ansible doesn’t have a module for working with qemu, or kvm. It does have a module for libvirt, which I found to be extremely lacking when looking into a virtualization solution. No cloud init, no image downloading. In addition to that, you have to define vm’s from xml templates if you want to make one, which is way too complex for what it does. If you’re doing all that work manually (or manuallyish, via templates), what’s the point of an automation solution?
Proxmox is not a complete replacement for VMware. Proxmox still does not have a distributed resource scheduler or distributed power management for it’s cluster which means the only time a VM will move between nodes is if a node goes down.
There’s no official support for VDI within proxmox and all the third party tools are janky at best, definitely not ready for enterprise level deployments.
Nvidia does not officially support vGPUs on proxmox. You can get it working but it’s definitely not something you’d want to run on production.
Proxmox is a decent option, or just use kvm provisioning directly with ansible.
So many companies can’t do this.
The staffing, the network and storage changes.
The suggestion to just use KVM and ansible is rather tone def.
Sounds like someone with limited experience in the industry, honestly.
Any shop large enough that this is such a massive undertaking is large enough that the people who care about this aren’t the people making financial decisions.
The good news is this is horrendous for finance as well so unless you cut a deal for your licensing costs because you’re a titan, you’ll be switching.
Actually, KVM and Ansible are a big part of the industry. KVM as a hypervisor is widely acknowledged in enterprise environments. Meanwhile, Ansible provides great automation for system management tasks. Their combined use for virtualization management and provisioning is not just common but a best practice, backed by a plethora of successful deployments across various industries.
I’m well aware of both, been in the industry for over 20 years.
But you still don’t seem to comprehend the cost or difficulty of the change.
Frankly the support options don’t seem very good either.
Deployment is not even half the battle, ongoing support is where the troubles really come out.
It’s also just dumb.
Ansible doesn’t have a module for working with qemu, or kvm. It does have a module for libvirt, which I found to be extremely lacking when looking into a virtualization solution. No cloud init, no image downloading. In addition to that, you have to define vm’s from xml templates if you want to make one, which is way too complex for what it does. If you’re doing all that work manually (or manuallyish, via templates), what’s the point of an automation solution?
Proxmox is not a complete replacement for VMware. Proxmox still does not have a distributed resource scheduler or distributed power management for it’s cluster which means the only time a VM will move between nodes is if a node goes down.
There’s no official support for VDI within proxmox and all the third party tools are janky at best, definitely not ready for enterprise level deployments.
Nvidia does not officially support vGPUs on proxmox. You can get it working but it’s definitely not something you’d want to run on production.