We recently had a customer have a datastore disappear due to a combo of a nic failure and switch misconfiguration and wanted to ask for opinions out there on some changes to possibly make. All of their vSwitches are vDS with load balancing based on NIC load. They do not want to create port channels on their switches so NIC load seems to be the best practice in this scenario. They have a single vmkernel port group for NFS (10.10.10.x/24) going to a NetApp. They have four vmnics in that vDS/port group with all active. They have three datastores on each controller (10.10.10.x/24). It is possible to assign multiple alias IPs on the virtual interface of the NetApp. So how would you provide high availablity and using all four (or multiple) vmnics actively? Should I create two more vmk's in different subnets (one for each datastore subnet), create aliases on the NetApp vifs and remount the datastores with different IPs? Any help will be appreciated and thanks in advance.
