As the use of virtualization expands, organizations face the challenges of network connectivity. Each new VM demands...
network access, potentially leading to network congestion, unacceptable latency, poor application performance and questionable application availability. While faster networking technologies are one alternative to network bandwidth, it's far more common and cost-effective to use network adapter teaming -- aggregating multiple inexpensive network ports to support faster and more resilient server connectivity. Hyper-V supports network interface card teaming under Windows Server 2016, and it can readily be employed to bolster network performance and resilience.
Hyper-V NIC teaming explained
NIC teaming is a technology that allows a computer to aggregate available physical network adapters installed in the system, such as enterprise-class servers. The aggregated NICs are abstracted into virtual network adapters, and those virtual network adapters can be provisioned to workloads in order to provide a combination of greater network bandwidth and higher resilience within the server.
For example, a server could aggregate four resident NICs into a team and present that additional bandwidth to a demanding workload to enhance application responsiveness or transaction capacity. At the same time, the additional NICs are still physically independent, so a fault in one NIC wouldn't cut off communication between the workload and clients or other back-office systems. The remaining NICs in the team would continue to carry the network traffic -- though total bandwidth would be reduced by the loss of that failed NIC.
Windows Server 2016 can aggregate between one and 32 physical Ethernet NICs. Windows Server 2016 also handles the algorithms needed to check NIC health and distribute network traffic between teamed NICs. It's theoretically possible to place a single NIC into a team, but this offers neither bandwidth nor fault tolerance benefits. This is done mainly for housekeeping or NIC management consistency. Administrators can manage Hyper-V NIC teaming through Windows PowerShell, Remote Desktop and Remote Server Administration Tools. This allows IT staff to configure and manage Hyper-V NIC teaming across the entire enterprise from a desktop.
Network resilience can be further improved by connecting the physical NIC port of each NIC team member to a different network switch -- not just a different port on the same switch. This extends fault tolerance to the switching architecture by eliminating a switch as a single point of failure. For example, if all four of the physical NIC ports above were connected to the same switch and that switch failed, the workload would become unavailable just as surely as if it were connected with a single NIC port. By distributing the team NIC ports to different switches -- this may require additional switches in the infrastructure -- the loss of one switch would not cut off connectivity to the server's workload.
Hyper-V NIC teaming limitations
Hyper-V NIC teaming in Windows Server 2016 is vendor-agnostic. It should support a wide variety of physical Ethernet NICs and should handle any NIC validated with Windows Hardware Qualification and Logo tests. However, there are some NICs that can't be used in an NIC team. For example, NICs used for network booting cannot be placed in a team. Teaming is currently an Ethernet-only technology, and non-Ethernet NIC technologies, such as Wi-Fi, Bluetooth and InfiniBand, aren't supported in NIC teams. NICs used for kernel debugging can't be used in NIC teams because these adapters are dedicated to remote debugging tasks. Finally, Hyper-V virtual NICs in the host partition cannot be teamed.
It's also important to note that some powerful NIC technologies won't work with Hyper-V NIC teaming under Windows Server 2016. Compatibility problems can occur when network data is exchanged directly with the NIC rather than passing data through the network stack where Hyper-V NIC teaming can intervene. For example, single-root I/O virtualization and TCP Chimney provide outstanding data offload capabilities that enhance network performance but aren't supported in Hyper-V NIC teaming because they bypass the OS.
Other technologies, like native quality of service (QoS) policies, can impose minimum bandwidth limitations, causing the NIC team performance to be less than it would be with QoS policies disabled. Finally, it's not possible to combine authentication with teaming, so technologies like 802.1X authentication can't be used in a NIC team. It's important to test Hyper-V NIC teaming in a lab or other protected environment to verify team performance and NIC hardware feature compatibility before implementing teaming in production.
Learn more about Windows Server 2016 networking capabilities
Hot add vNIC and modify memory in SCVMM 2016
Navigate Hyper-V files and folders
Related Q&A from Stephen J. Bigelow
Many compatibility issues can arise when moving VMs to the public cloud. Watch out for compatibility problems with partitions, OSes and image formats... Continue Reading
To migrate a VM and its dependencies from a local data center to a public cloud, use the forklift method to prepare the VM for migration, deploy the ... Continue Reading
Prepare your VMs with cloud migration best practices that examine how suitable a VM is for migration, what groundwork needs to be prepared for it and... Continue Reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.