SauRoNZA
Honorary Master
I do have another question for you guys - surely haven't a single chassis with multiple VMs is a single point of a failure? lightening, surge, PSU issues or even corrupt software knocks out all VMs?
Actually it's a benefit BECAUSE you only have one system to worry about.
Imagine you had to run 50 physical machines each with their own power supplies, hard drives, motherboards etc. You'd need to setup backup and replication and service routines for each and every one.
Instead now you have only one to worry about making things a lot simpler.
As Psywulf already said you can either run live replication to duplicate servers. You might have heard of a VMWare "Cluster" before, that implies more than one physical server split into multiple "nodes" that form a "Cluster" together. At any one time you can move VM's around from one physical machine to another as required (whether for redundancy failover or for processing power etc).
Normally this is couple with shared storage so the VM's containers (hard drives) all sit on one shared storage and the multiple nodes (cpu/motherboard/network etc) use the hard drive remotely meaning you can easily flip from on to the other system in real time. Therefore if one node goes down the others can pick up where that one left off without anyone actually noticing.
Then there are smaller setups where you maybe don't need such high redundancy or don't mind some down time. In those cases you can have two identical virtual servers using local storage and then simply run replication of the VM data every night for instance and should there be a catastrophic disaster you simply startup the VM on the other virtual server and absolutely nothing changes to the enduser experience.
The beauty of virtualisation is that because it's virtual if everything goes really pear shaped...let's say the building burns down but you manage to get the VM data out in time then you can simply go buy a new server and attach the storage to it and within a couple of hours you'll be back to normal again.
The same thing would be almost impossible if you have 50 physical servers instead.
*****
You can even flip from one Hypervisor to another with reasonable ease if the need arrises. So going from VMWare to HyperV to Proxmox is somewhat agnostic although not always 100% seamless.