Hello,
so:
With VSA, if a host in the cluster goes down, is there any downtime/dataloss?
In case of a host failure no dataloss, only a certain amount of downtime due to HA restarting the VMs to the surviving host(s)
My vendor says that if a host went down you would have to reboot all vm's that were running on the host that went down.
HA will do this automatically for you.
If I install VSA with 2 hosts, can I add a 3rd host months down the road? I've found a few blogs stating that you need all hosts ready at the time of install.
Correct, it's a static configuration you have to decide how many hosts will your VSA cluster have from the beginning.
How does the VSA use each ESXi's datastore(s). Do you setup each host with only 1 datastore? Or can each host have multiple datastores so you can have seperate raid arrays for performance and/or capacity?
Every host has one local datastore mapped on physical disks that are put in RAID 5/6/1+0. Every datastore is replicated to all other hosts so in case of host failure datastore survive. Every local storage is pooled to create an aggregate storage.
Regards,
Paolo.