Hello,
I'm looking for help in interpreting the results of tests on two neighbors. Please have a look at the test reports here attached.
My question is: why such different reports (many repetitions on one side) and what to do to minimize the repetitions that I suppose to slow down the throughput?
I'm looking for help in interpreting the results of tests on two neighbors. Please have a look at the test reports here attached.
My question is: why such different reports (many repetitions on one side) and what to do to minimize the repetitions that I suppose to slow down the throughput?
File Attachment:
By default iperf3 will take one throughput sample every second for 10 seconds, which is what your output shows. Finally it will summarize the stats from its 10 samples at the bottom of each report. Here is a link to the complete set of options for iperf3, and it may be interesting for you to experiment with different options. You can change the sample size by setting the -t (time) option, or you can specify the total number of bytes you want transmitted by specifying the -n (number of bytes) option instead of using a timed test. It may be helpful to set -O 3 to omit the first three seconds of the test which skips the TCP slow-start period.
The default settings for iperf3 has the client/sender initiating a TCP test. With TCP being connection-oriented all packets are expected to reach the destination, so checks are done for missing or corrupted packets and retransmitted if anything is amiss. On robust wired or data center networks you might expect a small number of retransmissions -- but over RF links of varying quality we would expect to see retransmissions. You could experiment by having iperf3 initiate a connectionless UDP test.
Thanks for answers.
"I’ll appreciate very much comments on my test reports with the two nodes connected DtD."
DtD through a switch or with a cross-over cable between POE injector's LAN ports?
Could be a bad cable or a bad switch.