I just ran into a (lab) situation where ESXi 5 hosts w/ 4 NICs, connected to Dell Equallogic arrays via two stacked switches, completely lost connectivity to the storage arrays when maintenance on the network was due. The hosts' NICs were equally spread over the switches, so one would not have expected unplanned outage in case a switch is rebooted.
So, who was the bad guy? The admin.
Turns out I forgot to add the Dell recommended "Storage Heartbeat" VMkernel port. Using the set-up script, add --heartbeat <IPinISCSIsubnet>, everything's fine now that the Equallogic group IPs can be reached via the newly configured VMkernel port which must not have a 1:1 vmk/vmnic binding.
For those of you who are to configure iSCSI on vSphere together with Equallogic arrays, this VMware KB article is worth reading. Follow the links for Dell configuration guides.
