NIC Teaming, also known as Windows Load Balancing or Failover (LBFO), is an extremely useful feature supported by Windows Server 2012 that allows the aggregation of multiple network interface cards to one or more virtual network adapters. This enables us to combine the bandwidth of every physical network card into the virtual network adapter, creating a single large network connection from the server to the network. Apart from the increased bandwidth, NIC Teaming offers additional advantages such as: Load balancing, redundant links to our network and failover capabilities.
Running Windows or Linux Server under ESX or Hyper-V? Get an award-winning backup solution for Free! Download Now!
Windows Hyper-V is also capable of taking advantage of NIC Teaming, which further increases the reliability of our virtualization infrastructure and the bandwidth available to our VMs.
Figure 1. Windows 2012 Server – Hyper-V NIC Teaming with Cisco Catalyst Switch
There are two basic NIC Teaming configurations: switch-independent teaming & switch-dependent teaming. Let’s take a look at each configuration and its advantages.
Switch-independent teaming offers the advantage of not requiring the switch to participate in the NIC Teaming process. Network cards from the server can connect to different switches within our network.
Switch-independent teaming is preferred when bandwidth isn’t an issue and we are mostly interested in creating a fault tolerant connection by placing a team member into standby mode so that when one network adapter or link fails, the standby network adapter automatically takes over. When a failed network adapter returns to its normal operating mode, the standby member will return to its standby status.
Switch-dependent teaming requires the switch to participate in the teaming process, during which Windows Server 2012 negotiates with the switch creating one virtual link that aggregates all physical network adapters’ bandwidth. For example, a server with four 1Gbps network cards can be configured to create a single 4Gbps connection to the network.
Switch-dependent teaming supports two different modes: Generic or Static Teaming (IEEE 802.3ad) and Link Aggregation Control Protocol Teaming (IEEE 802.1ax, LACP). LACP is the default mode in which Windows NIC Teaming always operates.
Load Balancing Mode - Load Distribution Algorithms
Load distribution algorithms are used to distribute outbound traffic amongst all available physical links, avoiding bottlenecks while at the same time utilizing all links. When configuring NIC Teaming in Windows Server 2012, we are required to select the required Load Balancing Mode that makes use of one of the following load distribution algorithms:
Hyper-V Switch Port: Used primarily when configuring NIC Teaming within a Hyper-V virtualized environment. When Virtual Machine Queues (VMQs) are used a queue can be placed on the specific network adapter where the traffic is expected to arrive thus providing greater flexibility in virtual environments.
Address Hashing: This algorithm creates a hash based on one of the characteristics listed below and then assigns it to available network adapters to efficiently load balance traffic:
- Source and Destination TCP ports plus Source and Destination IP addresses
- Source and Destination IP addresses only
- Source and Destination MAC addresses only
- Distributes outgoing traffic based on a hash of the TCP Ports and IP addresses with real-time rebalancing allowing flows to move backward and forward between networks adapters that are part of the same group.
- Inbound traffic is distributed similar to the Hyper-V port algorithm
Dynamic: The Dynamic algorithm combines the best aspects of the two previous algorithms to create an effective load balancing mechanism. Here’s what it does:
The Dynamic algorithm is the preferred Load Balancing Mode for Windows 2012 and the one we are covering in this article.
Click here for more teachnical articles covering Windows 2012 Server
Configuring NIC Teaming in Windows Server 2012