Are the tests meaningful?

[smadds]

This is more of a generic question. I have just set up a system to test the performance of a 350Mbit broadband link supplied by Virgin Media Business in the UK. I'm using a Nanopi Neo2 with gigabit ethernet port running headless Ubuntu 18.04 plugged straight into the Virgin router using a static public IP.

My issue is that there appear to be massive variations in performance - even through the night when the building is not being used and there must be little contention externally for the bandwidth. Is this typical of what others are seeing?

Also, I wondered whether there is away to lock the server that is being used so that this is not a variable in the testing?