For example:
NOTE: Changing the teaming to the image resolved this particular issue.
Scenario 1: The VM network portgroups are configured with all four uplinks set to "Active." However, uplinks 3 and 4 are dedicated strictly to storage traffic and are not trunked for virtual machine VLAN traffic. Following a vMotion, the virtual switch load balancing policy hashes the VM's MAC address to egress on either uplink 3 or 4, resulting in a Layer 2 traffic blackhole and immediate connectivity loss.
Scenario 2: Mixed combination of Active Unused with the incorrect ports so that 1 Data uplink is teamed with 1 VSAN uplink causing issues for the teaming algorithm and subsequently causing intermittent storage and VM network data loss.
This also explains why disconnecting and reconnecting the vNic occasionally restores connectivity, as this operation will initiate a new hash calculation for NIC teaming, potentially causing the VM to be mapped to an uplink configured for VM traffic.