Thanks for everybody who helped me to sort it out in the previous discussion. Here is the 'final picture' - let me know if I got it right.
1. HA
In the case of hardware failure HA will move VMs automatically to another ESX host but it will be done with a downtime. Guests are powered down then restarted onto another host in the cluster.
HA is what is monitoring the heartbeats of ESX hosts.
2. vMotion.
a) VMotion without DRS is only beneficial for moving VM's without downtime between hosts - and without DRS it must be initiated manually.
VMotion in itself does not detect a failure and perform a vMotion.
b) vMotion is independent of HA. That means if hardware failure occurs and vMotion is installed then there WILL BE a downtime. VMotion (and DRS): Move a VM between hosts only when both hosts are up and running
3. vMotion + DRS
vMotion + DRS will provide you load balancing capabilities. Machines will transparently flow between hosts if resource consumption exceeds the rules. But it won’t facilitate failover if hardware failure occurs.
4. FT
Fault Tolerance in combination with HA will provide you a transparent failover in the case of hardware failure.
