Clarification about VSAN stretched cluster

[Luca82]

Hi everyone, I'm struggling to understand the differences between fault domain and stretched cluster. Suppose there are 4 hosts in the same cluster, but 2 of them are in a location A and the other 2 are in the location B.... they are in the same network (same subnet)....this scenario can be imagined like in a big vessel: 2 hosts are in the fore of the vassel, and 2 in the aft... in the middle of the vessel....or maybe in the cloud, we can put the witness (virtual appliance) .....is this a stretched cluster??? So is it fair to say that the difference between the fault domain and the stretched cluster is the possibility to identify the witness in the latter case? Rather... if we setup without witness we must define 3 fault domain and VSAN automatically will choose on its own wich of them will be the witness? Is it correct?

[GreatWhiteTec · Accepted solution]

One way to think about it is...Fault domain is rack awareness, while stretched cluster is site awareness. Stretched cluster builds on FDs in the sense that we need at least 3 to have FTT=1 (2 copies of data, 1 witness object = 3).

The witness is needed for both 2-node and stretched clusters, given that you need 3 "locations" or Fault Domains to place objects. Witness object info is placed on the witness appliance, while the copies of data are placed on each node of the 2-node cluster or each site of the stretched cluster.

Normal vSAN clusters work a little different but I don't want to confuse you.

Here is a good document to help you out Storage and Availability Technical Documents