we do load-test acceptance testing using layer 4 distribution algo thanks to iperf3 running on different tcp ports.
start two servers (one per pipe)
on node1
( iperf3 -s & ); iperf3 -s -p 5202
failsafe test (single pipe)
on node2 or 3
iperf3 -c 10.5.5.1
multiple tcp/ports – now this should be about the double of the previous amount – for two pipes within the roughly balanced aggregate
on node2 or 3
( iperf3 -c 10.5.5.1 & ); iperf3 -c 10.5.5.1 -p 5202
load-test results are very similar across all tested modes,
however there’s a slight advantage with the balance-tlb mode (default dynamic setting disabled).