Intel 480T, User Manual

Page: 83 / 367

C H A P T E R 4 Configuring Switch Ports
81
logical port. For example, VLANs see the load-sharing group as a
single logical port. The algorithm also guarantees packet
sequencing between clients.
If a port in a load-sharing group fails, traffic is redistributed to the
remaining ports in the load-sharing group. If the failed port becomes
active again, traffic is redistributed to include that port. Load
sharing must be enabled on both ends of the link, or a network loop
will result.
This feature is supported between 480T switches only, but may be
compatible with third-party “trunking” or sharing algorithms.
Load Sharing Algorithms
Load sharing algorithms allow you to select the distribution
technique used by the load-sharing group to determine the output
port selection. Algorithm selection is not intended for use in
predictive traffic engineering.
You can configure one of three load-sharing algorithms on the
switch, as follows:
• Port-based — Uses the ingress port to determine which physical
port in the load-sharing group is used to forward traffic out of the
switch.
• Address-based — Uses addressing information to determine
which physical port in the load-sharing group to use for
forwarding traffic out of the switch. Addressing information is
based on the packet protocol, as follows:
- IP packets — Uses the source and destination MAC and IP
addresses, and the TCP port number.
- IPX packets — Uses the source and destination MAC
address, and IPX network identifiers.
- All other packets — Uses the source and destination MAC
address.
• Round-robin — When the switch receives a stream of packets, it
forwards one packet out of each physical port in the load-sharing
group using a round-robin scheme. Using the round-robin
algorithm, packet ordering is not guaranteed.