Hey cb122,
Welcome to the community and to Vmware,
Well first off HA is very different then FT. HA looks at the hosts and makes sure that all your hosts in your clusters are running and your VM's are up and running. If a host fails HA notices this and starts up all the VM's that was running on the failed host on another host in the cluster. Your VM's will see this as an unexpected shutdown. Although breif it can take a few mintues for all your VM's to boot again depending what was running on that failed host.
Fault Tolerance is essentially HA on stariods and what it does is eliminate that short outage or unexecpted shutdown on a host failure. However this comes at a high cost, reduces abilties, has high number of requirements to work ect ect, look at FT documentaiton for more information. FT's biggest draw back currently is your only allowed to have 1vCPU on a FT protected VM, which kind of shoots it in the foot as most systems that required such high availability require more vCPU's. However how FT works in short is this: You enable FT on a VM, it then creates a shadow copy of the VM and runs it on another host. All changes to the VM are always synced lived with its shadow copy. If a host fails that is running your VM, the shadow copy of the VM is then prompted to production and immedietly takes over all its functions, it then spawns off another shadow copy onto another host in the cluster to remain its protection.
To awenser your question every enviroment is different but for the most part a 3+ host Vmware cluster will protect you for most things, however in your case I would put a DRS rule in to split your 2 virtual DC's as if both are on one host and that host fails you will loose all AD and probably DNS as most people run these 2 services together. This could be problmatic to your enviroment.
Aside from that I would just split any VM's that couldn't handle having them all go down at once if a host failure occurs. A good example of this would be a web farm. Say you have 4 IIS servers in a web farm, you wouldn't want to run all 4 on one host as if that host fails your website goes down for 2-4 minutes until those web servers could start up on a new host. In which case you could create a DRS ruleset to keep 2 web servers on one host and 2 on another.
I hope this helps,
