[DBMandrake]

Hardware has issues with certain switches, so that lan needs to be throttled (RPi2 and such works on same switch)

^^ What is wrong stating these facts in this forum?

Sorry but this is very misleading to the point of not being true, so needs correcting.

RPi2 works fine on the same switch because it is only a 100Mbit Ethernet controller! Set the Vero interface to 100Mbit and it too will work fine on the same switch with broken flow control.

The RPi2 has a 100Mbit Ethernet controller attached to a 480Mbps USB2.0 hub, so it is impossible for the Ethernet controller to saturate the USB bus, so flow control is not required. The Vero has a Gigabit Ethernet controller attached to a 480Mbps USB2.0 hub so under the right conditions in Gigabit mode it is possible to saturate the USB bus and case receive overflows. Ethernet flow control solves the issue completely and is a standard feature on nearly all switches (including consumer grade) but there are a small number of switches that either don't have it or where it is broken or turned off.

The only time I have been able to reproduce the receive overflow issue was by deliberately turning off flow control on an enterprise switch. (Most consumer switches don't let you turn it off)

With a standard flow control enabled consumer grade cable modem/switch (my home cable router) I have measured iperf receive throughput of 466Mbps and send throughput of 350Mbps, with no packet loss. 466Mbps is very close to the theoretical maximum throughput of the USB bus.

While we're talking about problems, lets not forget the severe USB performance problems that plagued the Raspberry Pi for a long long time - as recently as late last year simply copying a file across the network using the built in Ethernet controller would cause a USB connected keyboard to suffer from missed keystrokes or "stuck down" key repeats because of lost USB packets. I personally experienced as many as 5 lost key strokes per 10 seconds, and a stuck down repeating key when typing at least every 30 seconds. It was so bad that I did not even attempt to type at the local console while heavy network activity was occurring.

The initial workaround for this was to drop to USB 1.1 mode, (great...) later on the USB driver was rewritten a couple of times to try to work around hardware deficiencies in the USB controller in the Pi. Although I have not tested it recently last time I checked this problem it was much improved but was not 100% solved and may never be, due to hardware limitations. I have never seen any problems with lost USB packets (USB keyboard lost keystrokes etc) on the Vero during high network activity.

Just so we don't give the impression here that the Raspberry Pi is without its own problems.