LACP setup on 1800-24G

[Craig LaHote]

Hello, have three servers, each running Broadcom BACS3. Have two NICs on each server setup as a team. Have the associated pair of ports on the 1800-24G setup as LACP, with a unique key for each pair. The switch shows all three pairs as aggregated on the LACP Status Overview screen, and the servers perform withour any problems using the teams.
However, the performance does not appear to be any different than just using a single NIC. Copying files between any of the three servers produces throughput rates essentially the same as copying to other similar servers with just a single NIC.
Am I missing something in the setup? I expected to see some improvement. Thanks.

[rick jones]

Aggregates do not increase the performance of a single "flow" - only of a collection of "flows." The definition of a "flow" depends on the packet scheduling algorithm being used.

A flow could be traffic between any two MAC addresses.

A flow could be traffic between any two IP addresses.

A flow could be traffic on a given TCP connection.

If your file transfer tests are a single flow - eg a single TCP connection, then it is expected that you would not see any increase in performance.

Now, on transmit, ie from your servers to the switch, it will be the software on the server that controls how things are spread-out. On receive, it will be the packet scheduling algorithms on the switch.

If you need a single flow to go faster you will have to upgrade to a faster underlying link technology.