I do a lot of wifi performance testing on airplanes and use testmy.net from time to time. I've noticed a change recently with your manual version of testmy.net for UL and DL testing. When I am in systems with small latencies, it seems to work fine, but for satellite-based systems where the intrinsic latency is 600 -1000 ms on one particular vendor, your test always comes back with a DL speed of between 1.7 -> 2.2 Mbps. These links we are testing are typically 10 -> 50 Mbps and so I know something is wrong. In the past it worked fine, but lately it seems like it clamps everything to about 1.7 Mbps. Last year, testmy.net did not have this issue and reported speeds similar to other speed tests. The problem is that not only I am using this but end customers are as well and I think the results could be misleading.
I run other tests on the planes like; fast.com, dslreports.com or our own tput tests and they come back with the expected data rates. I've also run our own tput tests using curl and file sizes up to 100 MBytes capturing packets with wireshark and am confident in the data rates we are seeing with the other tools. The data rates do vary over time due to link usage but each time I've tried testmy.net on the past 6 or so flights, its always giving me the same results.
Could you please look into this for me and let me know why this is acting this way?
synflood reacted to m_ahlenius in Issues with testmy.net and high latent environments