I would configure NFS as "jrmunday" described.
However.
I am reading the "Best Practices for Running VMware on Network-Attached Storage (NAS)" v2.0, January 2013
The multi-swithch configuration described there is on page 10 (see attached), whith "there are four network interface cards in the ESXi host, and these are configured in two pairs with IP hash failover. Each pair is configured as a team at their respective LAN switch". And the figure 7 depict the design, where EACH PAIR IS CONNECTED TO A SINGLE SWITCH, not a criss-cross pattern, as I would expect.
I may guess one pair connected to one switch because EtherChannel requires all links in a group to sit on one switch. Now, how does that design eliminates single point of failure??? If the switch goes down - so the connection from vmkport to that NFS storage.
That doesn't sound as a best practice to me. Unless I miss something..
Questions:
1) How is design from "Best Practices.. " better than two uplinks with active/standby teaming into two separate switches?
2) The other question I had - does anybody understand how the NFS is connected to the switch in that case? There supposed to be 4 connections from one storage to two switches. What are they - nic teams, independend?
Thank you