Are you using the latest version? VMware Infrastructure 3.5 does in fact protect against individual VM failure. So you are good to go.
VM Failure Monitoring, a.k.a. VM HA, allows you to direct a VM to be restarted on the same ESX server or to be VMotioned to another ESX server in the event that the VM fails. A heartbeat is maintained between VI and the VM to detect failure. To configure VM HA you must edit an advanced value in a cluster’s HA configuration screen. The PDF I linked shows you how to do that. Per the PDF, VM HA works like this:
- If no heartbeat is received from a virtual machine for 60 seconds or more, VMware HA resets that virtual machine.
- A virtual machine is allowed 240 seconds to stabilize after startup. This allows the virtual machine to perform a reset of the operating system and of VMware tools.
- If the virtual machine fails three or more times in a day, VMware HA does not reset the virtual machine.