Link to home
Start Free TrialLog in
Avatar of cmp119
cmp119Flag for United States of America

asked on

Windows Server 2012 NIC Team & Hyper-V

I have a Dell PowerEdge R720 server of which I setup a NIC Teaming.  This server has a total of four NICs, and I wanted to ensure the connection to the Hyper-V host stays connected at all time, so my thinking was to enable a NIC Team.  The other remaining NICs would be dedicated to two VMs.

First, I guess I need to know if configuring a NIC team on the host is recommended?

I am now installing Hyper-V Role, and during the process I selected the NIC Team adapter to Create Virtual Switches.  However, after installing Hype-V role I noticed it created an extra vitual adapter (vEthernet (Microsoft Network Adapter Multiplexor Driver - Virtual Switch).  I do not think we need virtual adapter since I have plenty of physical NICs, so unchecked the Virtual Switch Manager - Virtual Switch Properties - "Allow management operating system to share this network adapter".  After doing so, the server cannot access the network.  Now I cannot access the server via RDP, the server itself cannot access the network.  

If NIC Teams for the host is recommended, and works then how is Hyper-V installed and configured without the need of using a virtual adapter?
Avatar of Justin Yeung
Justin Yeung
Flag of United States of America image

Yes, the NIC teaming has to be done on the host level. Now, with Windows Server 2012 we have a choice of creating a team by using the manufacture driver or by using Windows Server 2012 teaming. I don’t have Windows Server 2012 teaming in production environment, so I cannot say which one is better.

Then, after the team is created, you create a virtual adapter (virtual switch) from the teamed adapter.

However, I would do the NIC setup a little bit differently:
- I will keep one NIC for host management, without creating a virtual switch from it
- I will team the 3 remaining NICs in Link Aggregation mode
- Create the virtual switch from the teamed adapter (3x 1Gbps)
- Disable the Management traffic on the virtual adapter

In that way you will have Hyper-V management traffic completely separated from the data from the virtual machines. And, the connection between virtual machines connected to the same virtual switch is theoretically 10Gbps.
Avatar of cmp119

ASKER

I am sorry for the severe delay getting back with you on this.  

I have abandoned nic teaming altogether and deciding going with one nic for the hypervisor.  There is something I am not understanding that I need clarification:

The server has a total of 4 physical NICs.

I want to use a NIC for the Hypervisor (VMHost).

One NIC for a Exchange 2010 VM (Exchange2010).  Will be conducting an Exchange 2003 to 2010 migration soon.

One NIC for an Exchange 2013 VM (Exchange2013) that will be created sometime in the future for Exchange 2010 to 2013 migration.

The last NIC (NIC1) was disabled since I do not need it for now.  

My first question is with Windows Server 2012, when installing the Hyper-V role it automatically created a virtual switch adapter.  I had all adapters disabled with the exception of (VMHost adapter).

I simply want to use the actual adapter for Hypervisor (Host).  

My question is the virtual switch adapter necessary for the hypervisror and each VM when I have physical adapter for all?  I spoke with a Dell rep and he stated it was needed and that's the only way it will work.  

I basically need a step-by-step instruction on how to setup the hypervisor nic properly, and then defining the nics for each installed vm.  I tried installing my first vm, and it would not display any physical nic that was enabled, and simply displayed Not connected and the virtual switch.

So, what I want to know exactly how to setup the adapters in a correct manner for the hypervisor host and then each vm.
HyperVNICAdapters.jpg
Yes, it is required.

Why did you abandon the teaming? As I said in my previous post, the best approach is to use one NIC for the Hyper-V host management traffic only and the other 3 in a team for traffic from all virtual machines.

Optionally, one physical NIC could be kept off for future expansion with iSCSI attached devices or for live migration between two Hyper-V hosts in a cluster, if ever used.

I will recommend against the approach of reserving a physical NIC per virtual machine, especially if there is communication going on between VMs on the same host.

Here is link to Windows Server 2012 Hyper-V Best Practices http://blogs.technet.com/b/askpfeplat/archive/2013/03/10/windows-server-2012-hyper-v-best-practices-in-easy-checklist-form.aspx and a video about configuring Hyper-V virtual switches in 2012 http://blogs.technet.com/b/keithmayer/archive/2012/11/26/configuring-hyper-v-virtual-networking-in-windows-server-2012-quot-early-experts-quot-challenge-exam-70-410-and-70-417.aspx
Avatar of cmp119

ASKER

I modified the configuration again as you suggested in that the Hypervisor NIC without creating a virtual switch for it.  

However, if I were to create a two NIC team via Windows Server 2012 teaming, how would would I be able to assign static IPs for each VM of which will use the same virtual switch.  

I presume we'll have one virtual switch for the team, and then I will be able to assign this same virtual switch to both VMs, but how will I be able to assign static IPs for each?
ASKER CERTIFIED SOLUTION
Avatar of Svet Paperov
Svet Paperov
Flag of Canada image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of cmp119

ASKER

Svet - I followed your advice.  All looks good!!!  Thank you.
My E-E Article: Some Hyper-V Software and Hardware Best Practices.

Please check out the article as it is based on building out Hyper-V standalone and cluster solutions since 2008 RTM.

We _always_ team unless the backend is a Scale-Out File Server cluster and we are using SMB Multi-Channel via 10GbE or higher with multiple NIC ports.

Why have a single point of failure, the physical NIC, when the goal is to create as much resilience to failure as possible?