Wednesday, July 10, 2013

vNetwork Load Balancing Policies

NIC Teaming
When we connect multiple physical Ethernet adapters (VMNICs) to a single virtual switch is called Teaming. A team can share the load of traffic between Physical and virtual networks among some or all of its members and provide passive failover in hardware failure or network outage.
Load Balancing
Load balancing spread network traffic from virtual machines on a virtual switch across two or more physical Ethernet adapters when we have NIC teaming.
Load Balancing Policies (Is Only for Outbound Traffic):
Route based on originating virtual port ID: This is the default configuration policy; in this policy Virtual ports of the vSwitch are associated with the physical Ethernet adapter.  The physical NIC is determined by the ID of the virtual port to which this virtual machine is connected. Traffic from a given virtual Ethernet adapter is consistently sent to the same physical adapter unless there is a failover to another adapter in the NIC team. This method is simple and fast and does not require the VMkernel to examine the frame for any information. This setting provides an even distribution of traffic if the number of virtual Ethernet adapters is greater than the number of physical adapters.
Replies are received on the same physical adapter as the physical switch learns the port association.
A VM cannot use more than one physical Ethernet adapter at any given time unless it has multiple virtual adapters.
Route based on source MAC hash:  In this policy, each VM’s outbound traffic is mapped to a specific physical NIC that is based on the virtual NIC’s MAC address. This method has low overhead and is compatible with all switches, but it might not spread traffic evenly across the physical NICs.
Choose an uplink based on a hash of the source Ethernet MAC address (VM virtual NIC’s MAC Hash). Traffic from a given virtual NIC is consistently sent to the same physical adapter unless there is a failover to another adapter in the NIC team. This setting provides an even distribution of traffic if the number of virtual Ethernet adapters is greater than the number of physical adapters.
Replies are received on the same physical adapter as the physical switch learns the port association.
A given virtual machine cannot use more than one physical Ethernet adapter at any given time unless it uses multiple source MAC addresses for traffic it sends.
Route based on IP hash: In this policy, a physical NIC for each outbound packet (packet send from VM) is chosen based on its source and destination IP address. Selection of uplink will be based on a hash of the source and destination IP addresses of each packet. (For non-IP packets, whatever is at those offsets is used to compute the hash.) This method has higher CPU overhead but a better distribution of traffic across physical NICs associated with vSwitch.
You can use link aggregation; grouping multiple physical adapters to create a fast network pipe for a single virtual adapter in a virtual machine. All adapters in the NIC team must be attached to the same physical switch or an appropriate set of stacked physical switches. 802.3ad link aggregation support or EtherChannel must be supported on the physical switch. The Link Aggregation Control Protocol is a method to control the bundling of several physical ports to form a single logical channel. (LACP is part of the IEEE 802.3ad specification.) EtherChannel and IEEE 802.3ad standards are similar and accomplish the same goal. EtherChannel is a port trunking technology used primarily on Cisco switches. This technology allows grouping several physical Ethernet links to create one logical Ethernet link for providing fault tolerance and high-speed links between switches, routers, and servers. When one VM communicates to different Physical Machines, it chooses different physical NICs. On the return traffic, the packet can come in on multiple paths because more than two NICs might be teamed. Thus, link aggregation must be supported on the physical switch.
A single virtual NIC of a virtual machine might use the bandwidth of multiple physical adapters associated with vSwitch.
The physical switch sees the client MAC address on multiple ports. There is no way to predict which physical Ethernet adapter will receive inbound traffic. So the link aggregation must be supported on the physical switch.
Thanks to VMware, Information is from the white paper provided by VMware.
    

0 comments:

Post a Comment