Dual Channel Redundancy Mechanisms; Basic Redundancy Mechanisms; Advanced Redundancy Mechanisms; Dual Channel Network Adapter Failure Recovery - HP Rx2620-2 - Integrity - 0 MB RAM Manual

The client's ARP table entries expire after a few minutes; as a result, the clients will re-ARP
for the server's IP address. The clients may or may not get the same IP address for the server
depending on the results from the load-balancing algorithm.
• Periodic non-cross trunk load balancing when server transmits ARP request:
Periodically, the server must broadcast an ARP request for the MAC address of an IP address
to which it needs to talk. Whenever the server broadcasts the ARP request, all clients within
the broadcast domain receive it. Any client that has the server's IP address in its ARP table
may use the Source MAC Address (one of the server's MAC addresses from one of the
groups) in the ARP data to update its ARP table. If most, or all, of the clients do this, then
all receive traffic to the server will be received in a single group instead of in both groups.
This happens because all clients have the same MAC address in their ARP table for the
server. This condition will eventually correct itself as soon as client ARP table entries begin
to expire. As the client's re-ARP for the server's MAC address, receive load balancing will
slowly spread across both groups in the Dual Channel team.
It is possible to speed up the recovery process by reducing the ARP table timeout value in
the clients IP stack. Refer to the OS manufacturer for information on how to accomplish this.

Dual Channel Redundancy Mechanisms

Basic Redundancy Mechanisms

The basic redundancy mechanisms on Dual Channel work exactly like that of NFT or TLB with
one exception – the heartbeat mechanisms treat all teamed ports within a group as a single teamed
port. In other words, when using receive validation heartbeats, the mechanism need only receive
a frame on any teamed port in the group to validate the entire group for the mechanism.
This is done since teaming cannot predict how the switch load-balancing algorithm will deliver
frames to an individual teamed port within a group.

Advanced Redundancy Mechanisms

The advanced redundancy mechanisms on Dual Channel work exactly like that of NFT or TLB
with one exception – the Active Path and Fast Path mechanisms treat all teamed ports within a
group as a single teamed port. In other words, when using Active Path, the mechanism need
only receive an Echo Node response frame on any teamed port in the group to validate the entire
group for the Active Path mechanism. In addition, Fast Path needs only to receive a Spanning
Tree configuration BPDU on at least one teamed port in the group to validate connectivity with
the preferred root switch and to choose the group with the higher bandwidth path to the root
switch.
This is done since teaming cannot predict how the switch load-balancing algorithm will deliver
frames to an individual teamed port within a group.

Dual Channel Network Adapter Failure Recovery

Network adapter failure recovery for Dual Channel is identical to SLB within a particular group.
Failure recovery between groups is identical to TLB.

Dual Channel Applications

Dual Channel is ideal for environments that need NIC fault tolerance, switch fault tolerance,
transmit load balancing and receive load balancing, all in a single team type.

Recommended Configurations for Dual Channel Environments

HP recommends the following configurations for Dual Channel teams:
78 The Mechanics of Teaming for the Advanced User