Plain Virtualization

You might not be aware as most customer do not have a datastore go beyond 2PB. Today we discover that if a datastore on vVOLs or vSAN which is more than 2PB, it will not be able to be use to create any workload. This is due to the PlaceVm API which limits it. You can refer to this kb . Though this will not affect many customers, VMware is committed to have this worked on perhaps to remove this limit. While that is at work, it is good to know and be aware when architecting.

Some people asked me this question: Will HA be activated if datastore is lost on a host due to iSCSI connection failure or FC connection on just one host without losing the host server from the management network? Though I know the answer however I was curious myself and decide to find out by performing this on a iSCSI connection to see the results.   From my understanding, HA protects hosts failure however as long the host primary detection method (management network) is still alive, the host is considered alive in such, even though the VMs that the host is hosting is "gone" due to perhaps a nic failure for iSCSI connection or a HBA failure for FC connection, HA will not be activated even though datastore heartbeat (only in vSphere 5.x and its a secondary detection if primary failed) is lost. I realize vCenter will initialed a migration of the VM however this will get the process halt.  We have to kill the process and wait for a timeout in order to take...