Ideally, if you are using network attached storage you would want at least 6 nics (at 1Gb - if you are using 10G things may be a bit different...) - 2 for management, 2 for storage traffic and 2 for workload traffic. Now with 4 it will depend on your usage patterns. If one NIC is sufficient bandwidth for your storage and your workloads I would suggest something like this:
i) two nics for management - one primary for service console access, and the other primary for vMotion. With the other one configured as stand-by in each case.
ii) two nics for storage and workloadds - again one primary for storage, and the other primary for workloads traffic. With the other one configured as stand-by in each case.
This way your storage and workloads traffic will not compete for bandwidth under normal conditions, only in case of failure will you have two types of traffic traversing the same link...
You may need to change the combination if your usage requires it, but just remember that with HA it is very important to have your service console connected redundantly, so you don't trigger HA failover of all the VMs in case the SC looses connectivity.
Cheers,
Milos
