In several cases, it's not clear what the actual (not theoretical) bandwidth available to each section of the lab is.
I think we should PXE boot to SLAX, and use iperf to measure the bandwidth. (with our server operating as the iperf server and the local computer acting as the iperf client)