Windows 2012 NIC Teaming

Windows Server 2012 supports up to 32 network cards in a "dual live" mode, which increases throughput by 32 times times.

Group mode:

switch-independent (Switch independent):

This is the default when configured, this mode does not require the switch to participate in the combination configuration, because the switch in standalone mode does not know that the NIC is a part of the host,the NIC in the teaming group can be connected to a different switch .

In addition, the main standby mode is used when connecting the different switches, and the load balancer aggregation can be achieved only when the connection is on the same switch .

1) switch-independent/ Address Hash distribution

This configuration will send packets to all active NIC members via the selected address hash, because at the time of routing, an IP address can only be associated with one MAC address, so this method can only be received by a team member when communicating with the revenue station. This also means that inbound traffic does not exceed the bandwidth of one member of the team. Most suitable for use in the following situations:
You need to consider making team on different physical switches, using Active/Standby (active/standby) mode, Team VMS, running workflows on the server with a lot of outbound load, and a small amount of inbound load, such as IIS.

2) Switch independent/hyper-v Port Distribution

This approach distributes traffic across all active team members (NICs) based on the port number of the Hyper-V switch. The bandwidth for each Hyper-V port does not exceed the bandwidth of one team member (NIC).
Because the Hyper-V port number is bound to the team NIC, both inbound and outbound traffic on the same VM is using the same team NIC.
This configuration method is best used in the following situations (also):
The number of VMS is much larger than the number of team members, and the bandwidth that can be accepted for a VM is limited to the available bandwidth of a team NIC member.

switch-dependent (Switch dependent):

This mode requires that the NIC in the teaming group be connected to the same switch ( or to a single physical switch implemented as a CASCADE multi-switch mode), and is divided into the following two types of modes supported by the switch:

1, Static Group (IEEE 802.3ad Draft v1): This mode configures the type of link between the switch and the host, since this is a statically configured solution without any additional protocol, Therefore, it will not cause the combination of the switch or host due to cable insertion errors or other errors. In this mode, the network card can work at different speeds, that is, can be used to create a combination of non-speed network cards, but also requires the switch fully support IEEE 802.3AD standard, generally, server-level switches generally support this mode.

2,Laca dynamic combination (IEEE 802.1ax, LACP): Laca dynamic combination is a link aggregation to the same switch, but not statically configured, but dynamically constituted (that is, autonegotiation). It is implemented through an intelligent Link negotiation protocol, LACP (link Aggregation Control Protocol). LACP was originally used for link aggregation between switches and switches, and the LACP protocol enabled 2 switches to send LACP negotiation messages to each other, and when they found that there were multiple links available between the 2, the links were automatically combined into a larger, more bandwidth logical link. Thus, the purpose of widening the link bandwidth between switches is realized by using load balancing.

1) Switch dependent/address Hash Distribution
This configuration sends packets to all active NIC members via the selected address hash. Like all switch dependent configurations, the switch decides how to distribute inbound traffic on the team NIC.
This configuration is best used in the following cases (at the same time):
Maximum performance is achieved when the team is guaranteed to use the same switch, and the bandwidth required to do TEAMING,VM under the Hyper-V switch needs to be greater than the available bandwidth of a team NIC.

2) switch dependent/hyper-v port distribution
This method is based on the port number of the Hyper-V switch to all active team Distribute traffic on members (NICs). The bandwidth for each Hyper-V port does not exceed the bandwidth of one team member (NIC). Like all switch dependent configurations, the switch decides how to distribute inbound traffic on the team NIC.
This configuration is best used for the following scenarios (at the same time):
The number of VMS is much larger than the number of team NICs, requires a team approach with a specific protocol (such as LACP), and limits the bandwidth of a VM to a NIC in a team member.

650) this.width=650; "Src=" http://s3.51cto.com/wyfs02/M00/45/52/wKiom1PnKoihduYqAAFMjlEHMDE176.jpg " title= "QQ picture 20140810161539.jpg" alt= "wkiom1pnkoihduyqaafmjlehmde176.jpg"/>