BACKGROUND:

We recently got MS Hyper-V and and are wanting to create some virtual servers that will be connected to the SAN. We have a CX3-10 DELL/EMC SAN. When using a physical server, they have 2 physical NIC's with 2 ports each on them. One NIC for LAN and one NIC that is dedicated for iSCSI SAN use only (on both ports). One port is for a 0.x subnet leading to a DELL switch. The other port is for a 1.x subnet leading to a 2nd DELL switch. We also use PowerPath for redundancy. The other LAN NIC has both of its ports teamed to create a single virtual NIC. That is how we do physical servers.

QUESTION:

The hardware of our Hyper-V server is newer and it has a single onboard NIC with 4 ports and then a PCI slot NIC with 4 ports of its own. The PCI NIC will be dedicated for SAN use only, but I am trying to figure out the optimum way to make use of all 4 ports. And if there is even any benefit (in the form of higher I/O throughput, etc.) to using all 4 physical ports or just 2? This is an Intel gigabit card we are talking about BTW. So there are several scenarios I came up with and would like to know which is best and why/whynot? Right now, the plan is to only have one virtual server (guest) that is connected to the SAN which we will call FILESERVER for purposes of this question. It is a file server that contains tons of data utilized by 600 nodes or so.

1.) on the Hyper-V physical server itself, team these 4 SAN ports into 2 Virtual adapters, and then present these 2 virtual adapters as SAN0 and SAN1 respectively, to the FILESERVER guest machine, for connection to the 2 redundant DELL switches (0.X & 1.X iSCSI subnets).

2.) present only 2 physical ports to the FILESERVER guest machine and set it up just like our existing physical servers - no real change in how we are already hooking up hosts to our SAN

3.) present all 4 ports to the FILESERVER guest machine (unteamed) and then use all 4 connections for SAN traffic. WOULD THIS EVEN WORK? Since we only have 2 iSCSI subnets and switches, then I don't see any benefit to this or if it is even a possibility. But management wanted me to find out if it was possible in case it would somehow increase I/O ability and/or redundancy. If this would even work with only 2 iSCSI subnets, how would it affect Power Path. etc...I am hoping that option 1 or 2 is gonna be an obvious solution once you experts read this

OTHER:

Is this a scenario where PowerPath VE is warranted??? I don't know anything about it (yet) but I expect it is really for more complex situations and maybe even when using VHD's that boot from the SAN itself (which we don't do)

Let me know if you need clarification on anything and thanks in advance!