Some specific posts made by a FileZilla developer piqued my interest. See here and here.
I remember reading a thread here recently discussing how theoretically impossible it was for a single TCP stream to go faster than a particular speed due to constraints like latency among other things.
I would imagine that the developer was likely testing in perfect conditions, a completely uncontested link both locally and at the destination as well as good routing in between but the distance was still transatlantic (20+ hops)...
Has anyone here seen speeds like that using a single TCP thread of any kind, not just FTP?
The best I have achieved is with sftp and was ~2.9MiB/s. Is this developer smoking something odd or is he correct that there is an underlying problem with congestion globally?
A well routed VPS I test with based in the US is roughly 10 hops from me and even taking in to account an ISP with plenty of capacity and a good home network configuration there is no hope I would even reach 2Mbit/s for a single stream via FTP/sftp/HTTP.