Looking for automatic failover when a machine dies - vSphere HA / vMotion / Replication don't appear to satisfy this need

[drzap]

I'd like to understand if there is a way to automatically fail over to another VM on another machine in case of an *unexpected* machine crash or power outage.  It needs to include the state of the machine.  Choices I've found:

1. HA look like it requires a restart.
2. Replication also appears to involve a VM restart.  I haven't determined the difference between HA & Replication yet.
3. vMotion - The documents are not clear about supporting unplanned migrations.  I'm seeing one document claims a 2-second live migration using vMotion - but no other.  However, vMotion is limited by how many VMs can be restarted at once.  If a machine goes down you don't have ANY time to queue up migrations.

Does anyone know if unexpected migrations can be supported; and if so, any pointers to (clear) documentation?

Thanks.

[daphnissov]

Fault Tolerance may be what you want.

[drzap]

Thanks daphnissov.  To tie this off, the document that covers this is called vSphere Availability.  It describes solutions for both "slow" stateless recovery via restarts, and two solutions for automatic stateful failover (vSphere Fault Tolerance and vCenter HA).

[sjesse]

Just so its clear vcenter ha is only for the vcenter vm, it won't work for others.

[daphnissov]

Yes, so what you want is Fault Tolerance not, as sjesse​ pointed out, vCenter HA.